vecId
stringlengths
12
23
id
stringlengths
2
13
conference
stringclasses
11 values
year
float64
2.02k
2.03k
title
stringlengths
6
189
abstract
stringlengths
10
4.74k
author
stringlengths
0
7.45k
aff
stringlengths
0
7.16k
status
stringclasses
11 values
track
stringclasses
4 values
keywords
stringlengths
0
804
github
stringlengths
0
141
site
stringlengths
0
193
gsCitation
float64
-1
11.1k
arxiv
stringlengths
0
12
text
stringlengths
58
4.82k
vector
list
corl_2023_VLihM67Wdi6
VLihM67Wdi6
corl
2,023
STERLING: Self-Supervised Terrain Representation Learning from Unconstrained Robot Experience
Terrain awareness, i.e., the ability to identify and distinguish different types of terrain, is a critical ability that robots must have to succeed at autonomous off-road navigation. Current approaches that provide robots with this awareness either rely on labeled data which is expensive to collect, engineered features...
Haresh Karnan;Elvin Yang;Daniel Farkash;Garrett Warnell;Joydeep Biswas;Peter Stone
University of Texas, Austin;University of Texas at Austin;;Army Research Laboratory;The University of Texas at Austin;University of Texas, Austin
Poster
main
Vision-based Navigation;Representation Learning;Learning from Experience
https://github.com/HareshKarnan/sterling_corl23
https://openreview.net/forum?id=VLihM67Wdi6
23
STERLING: Self-Supervised Terrain Representation Learning from Unconstrained Robot Experience Terrain awareness, i.e., the ability to identify and distinguish different types of terrain, is a critical ability that robots must have to succeed at autonomous off-road navigation. Current approaches that provide robots with...
[ -0.06970109045505524, -0.03929482027888298, -0.001080977963283658, 0.04285024479031563, -0.006374760530889034, -0.01342542190104723, 0.011943995952606201, -0.002153854351490736, -0.004108642227947712, 0.018203020095825195, -0.010425534099340439, -0.007495088968425989, 0.01914742961525917, ...
corl_2023_VscdYkKgwdH
VscdYkKgwdH
corl
2,023
Neural Graph Control Barrier Functions Guided Distributed Collision-avoidance Multi-agent Control
We consider the problem of designing distributed collision-avoidance multi-agent control in large-scale environments with potentially moving obstacles, where a large number of agents are required to maintain safety using only local information and reach their goals. This paper addresses the problem of collision avoidan...
Songyuan Zhang;Kunal Garg;Chuchu Fan
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Distributed control;Control barrier functions;Graph neural networks
https://openreview.net/forum?id=VscdYkKgwdH
33
Neural Graph Control Barrier Functions Guided Distributed Collision-avoidance Multi-agent Control We consider the problem of designing distributed collision-avoidance multi-agent control in large-scale environments with potentially moving obstacles, where a large number of agents are required to maintain safety using o...
[ -0.009806978516280651, -0.04492582380771637, -0.03261687979102135, -0.027704257518053055, 0.0052596088498830795, -0.005866837687790394, -0.012948133982717991, -0.003549779299646616, 0.04518149793148041, 0.034936219453811646, -0.03900876268744469, -0.03647027164697647, -0.004650096874684095, ...
corl_2023_VtJqMs9ig20
VtJqMs9ig20
corl
2,023
CAT: Closed-loop Adversarial Training for Safe End-to-End Driving
Driving safety is a top priority for autonomous vehicles. Orthogonal to prior work handling accident-prone traffic events by algorithm designs at the policy level, we investigate a \textbf{C}losed-loop \textbf{A}dversarial \textbf{T}raining (CAT) framework for safe end-to-end driving in this paper through the lens of e...
Linrui Zhang;Zhenghao Peng;Quanyi Li;Bolei Zhou
Tsinghua University;University of California, Los Angeles;Shanghai Artificial Intelligence Laboratory;University of California, Los Angeles
Poster
main
Safety-Critical Scenario Generation;Adversarial Training;End-to-End Driving
https://github.com/metadriverse/cat
https://openreview.net/forum?id=VtJqMs9ig20
33
CAT: Closed-loop Adversarial Training for Safe End-to-End Driving Driving safety is a top priority for autonomous vehicles. Orthogonal to prior work handling accident-prone traffic events by algorithm designs at the policy level, we investigate a \textbf{C}losed-loop \textbf{A}dversarial \textbf{T}raining (CAT) framewo...
[ -0.05890648066997528, -0.02158306911587715, -0.014708188362419605, 0.007731936406344175, -0.03190460428595543, -0.02147248014807701, -0.020845815539360046, 0.027646970003843307, -0.015086030587553978, 0.03081715665757656, -0.0018742342945188284, -0.03295518830418587, -0.02156463824212551, ...
corl_2023_VtUZ4VGPns
VtUZ4VGPns
corl
2,023
IIFL: Implicit Interactive Fleet Learning from Heterogeneous Human Supervisors
Imitation learning has been applied to a range of robotic tasks, but can struggle when robots encounter edge cases that are not represented in the training data (i.e., distribution shift). Interactive fleet learning (IFL) mitigates distribution shift by allowing robots to access remote human supervisors during task exe...
Gaurav Datta;Ryan Hoque;Anrui Gu;Eugen Solowjow;Ken Goldberg
University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;;University of California, Berkeley
Poster
main
Imitation Learning;Fleet Learning;Energy-Based Models
https://github.com/BerkeleyAutomation/IIFL
https://openreview.net/forum?id=VtUZ4VGPns
5
IIFL: Implicit Interactive Fleet Learning from Heterogeneous Human Supervisors Imitation learning has been applied to a range of robotic tasks, but can struggle when robots encounter edge cases that are not represented in the training data (i.e., distribution shift). Interactive fleet learning (IFL) mitigates distribut...
[ -0.07951296865940094, -0.047066304832696915, -0.01360335573554039, 0.014283989556133747, -0.02377557009458542, -0.004918279591947794, 0.031756702810525894, -0.01976635679602623, 0.002405528211966157, 0.005943892523646355, -0.022880490869283676, -0.027038883417844772, -0.033062029629945755, ...
corl_2023_W0zgY2mBTA8
W0zgY2mBTA8
corl
2,023
ChainedDiffuser: Unifying Trajectory Diffusion and Keypose Prediction for Robotic Manipulation
We present ChainedDiffuser, a policy architecture that unifies action keypose prediction and trajectory diffusion generation for learning robot manipulation from demonstrations. Our main innovation is to use a global transformer-based action predictor to predict actions at keyframes, a task that requires multi- modal s...
Zhou Xian;Nikolaos Gkanatsios;Theophile Gervet;Tsung-Wei Ke;Katerina Fragkiadaki
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Manipulation;Imitation Learning;Transformers;Diffusion Models
https://github.com/zhouxian/chained-diffuser
https://openreview.net/forum?id=W0zgY2mBTA8
88
ChainedDiffuser: Unifying Trajectory Diffusion and Keypose Prediction for Robotic Manipulation We present ChainedDiffuser, a policy architecture that unifies action keypose prediction and trajectory diffusion generation for learning robot manipulation from demonstrations. Our main innovation is to use a global transfor...
[ -0.048928260803222656, -0.03552648797631264, -0.04385045915842056, 0.010717791505157948, -0.039534326642751694, 0.029705148190259933, -0.05041533336043358, 0.026477117091417313, -0.0030126788187772036, 0.05607345700263977, -0.009257923811674118, -0.046208009123802185, -0.018180349841713905, ...
corl_2023_W5SrUCN0yUa
W5SrUCN0yUa
corl
2,023
A Bayesian Approach to Robust Inverse Reinforcement Learning
We consider a Bayesian approach to offline model-based inverse reinforcement learning (IRL). The proposed framework differs from existing offline model-based IRL approaches by performing simultaneous estimation of the expert's reward function and subjective model of environment dynamics. We make use of a class of prior...
Ran Wei;Siliang Zeng;Chenliang Li;Alfredo Garcia;Anthony D McDonald;Mingyi Hong
Texas A&M University - College Station;University of Minnesota, Twin Cities;The Chinese University of Hong Kong;Texas A&M University - College Station;University of Wisconsin - Madison;University of Minnesota, Minneapolis
Poster
main
Inverse Reinforcement Learning;Bayesian Inference;Robustness
https://github.com/rw422scarlet/bmirl_tf
https://openreview.net/forum?id=W5SrUCN0yUa
7
A Bayesian Approach to Robust Inverse Reinforcement Learning We consider a Bayesian approach to offline model-based inverse reinforcement learning (IRL). The proposed framework differs from existing offline model-based IRL approaches by performing simultaneous estimation of the expert's reward function and subjective m...
[ -0.07442260533571243, 0.015275420621037483, 0.00654660863801837, 0.00808838289231062, -0.008031455799937248, -0.035370662808418274, 0.04201215133070946, 0.023909352719783783, 0.024763258174061775, 0.0275716595351696, -0.061708904802799225, 0.005820788908749819, -0.02679365687072277, 0.0194...
corl_2023_W7eg2NqFJ60
W7eg2NqFJ60
corl
2,023
Transforming a Quadruped into a Guide Robot for the Visually Impaired: Formalizing Wayfinding, Interaction Modeling, and Safety Mechanism
This paper explores the principles for transforming a quadrupedal robot into a guide robot for individuals with visual impairments. A guide robot has great potential to resolve the limited availability of guide animals that are accessible to only two to three percent of the potential blind or visually impaired (BVI) us...
J. Taery Kim;Wenhao Yu;Yash Kothari;Bruce Walker;Jie Tan;Greg Turk;Sehoon Ha
Google;Georgia Institute of Technology;Georgia Institute of Technology;Google;;Georgia Institute of Technology;Georgia Institute of Technology
Poster
main
Assistive Robot;Autonomous Navigation;Interaction Modeling
https://openreview.net/forum?id=W7eg2NqFJ60
13
Transforming a Quadruped into a Guide Robot for the Visually Impaired: Formalizing Wayfinding, Interaction Modeling, and Safety Mechanism This paper explores the principles for transforming a quadrupedal robot into a guide robot for individuals with visual impairments. A guide robot has great potential to resolve the l...
[ -0.025009561330080032, -0.01036818977445364, 0.00118573650252074, 0.02779475972056389, -0.039565086364746094, -0.00844621192663908, 0.03347961977124214, 0.013935533352196217, 0.028233524411916733, 0.018590249121189117, -0.014431527815759182, -0.005026712082326412, 0.03887832537293434, -0.0...
corl_2023_W8MjsxHrDpL
W8MjsxHrDpL
corl
2,023
Synthesizing Navigation Abstractions for Planning with Portable Manipulation Skills
We address the problem of efficiently learning high-level abstractions for task-level robot planning. Existing approaches require large amounts of data and fail to generalize learned abstractions to new environments. To address this, we propose to exploit the independence between spatial and non-spatial state variabl...
Eric Rosen;Steven James;Sergio Orozco;Vedant Gupta;Max Merlin;Stefanie Tellex;George Konidaris
Brown University;University of the Witwatersrand;Brown University;Brown University;Brown University;, Brown University;Brown University
Poster
main
Learning Abstractions;Mobile Manipulation
https://github.com/ericrosenbrown/aosm_experiments
https://openreview.net/forum?id=W8MjsxHrDpL
5
Synthesizing Navigation Abstractions for Planning with Portable Manipulation Skills We address the problem of efficiently learning high-level abstractions for task-level robot planning. Existing approaches require large amounts of data and fail to generalize learned abstractions to new environments. To address this, ...
[ -0.034834396094083786, -0.030344383791089058, -0.008446374908089638, 0.03625132516026497, -0.007397478446364403, -0.012476344592869282, -0.0014169299975037575, 0.0013490737183019519, 0.028283394873142242, 0.040115680545568466, -0.030657213181257248, -0.028228189796209335, -0.0036550352815538...
corl_2023_WGSR7HDuHu
WGSR7HDuHu
corl
2,023
Learning Robot Manipulation from Cross-Morphology Demonstration
Some Learning from Demonstrations (LfD) methods handle small mismatches in the action spaces of the teacher and student. Here we address the casewhere the teacher’s morphology is substantially different from that of the student. Our framework, Morphological Adaptation in Imitation Learning (MAIL), bridges this gap allo...
Gautam Salhotra;I-Chun Arthur Liu;Gaurav S. Sukhatme
;University of Southern California;University of Southern California
Poster
main
Imitation from Observation;Learning from Demonstration
https://github.com/uscresl/mail
https://openreview.net/forum?id=WGSR7HDuHu
8
Learning Robot Manipulation from Cross-Morphology Demonstration Some Learning from Demonstrations (LfD) methods handle small mismatches in the action spaces of the teacher and student. Here we address the casewhere the teacher’s morphology is substantially different from that of the student. Our framework, Morphologica...
[ -0.08107157051563263, -0.052128054201602936, 0.007542405277490616, -0.0258039440959692, -0.019227560609579086, -0.017945723608136177, 0.011908083222806454, -0.010812019929289818, 0.0035691740922629833, 0.006910775322467089, -0.046480536460876465, -0.011694444343447685, -0.0076306480914354324...
corl_2023_WWiKBdcpNd
WWiKBdcpNd
corl
2,023
HANDLOOM: Learned Tracing of One-Dimensional Objects for Inspection and Manipulation
Tracing – estimating the spatial state of – long deformable linear objects such as cables, threads, hoses, or ropes, is useful for a broad range of tasks in homes, retail, factories, construction, transportation, and healthcare. For long deformable linear objects (DLOs or simply cables) with many (over 25) crossings, w...
Vainavi Viswanath;Kaushik Shivakumar;Mallika Parulekar;Jainil Ajmera;Justin Kerr;Jeffrey Ichnowski;Richard Cheng;Thomas Kollar;Ken Goldberg
University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Carnegie Mellon University;Toyota Research Institute;Toyota Research Institute;University of California, Berkeley
Oral
main
state estimation;deformable manipulation
https://github.com/vainaviv/handloom
https://openreview.net/forum?id=WWiKBdcpNd
6
HANDLOOM: Learned Tracing of One-Dimensional Objects for Inspection and Manipulation Tracing – estimating the spatial state of – long deformable linear objects such as cables, threads, hoses, or ropes, is useful for a broad range of tasks in homes, retail, factories, construction, transportation, and healthcare. For lo...
[ -0.013746781274676323, -0.013160642236471176, -0.021503958851099014, -0.00506460340693593, -0.010605445131659508, 0.009236262179911137, -0.004922647960484028, -0.006754331290721893, -0.03742128610610962, 0.02124752290546894, -0.01730024628341198, -0.0047806925140321255, 0.049455441534519196,...
corl_2023_WmF-fagWdD
WmF-fagWdD
corl
2,023
SCALE: Causal Learning and Discovery of Robot Manipulation Skills using Simulation
We propose SCALE, an approach for discovering and learning a diverse set of interpretable robot skills from a limited dataset. Rather than learning a single skill which may fail to capture all the modes in the data, we first identify the different modes via causal reasoning and learn a separate skill for each of them. ...
Tabitha Edith Lee;Shivam Vats;Siddharth Girdhar;Oliver Kroemer
Carnegie Mellon University;;Carnegie Mellon University;Carnegie Mellon University
Poster
main
skill discovery;causal learning;manipulation
https://openreview.net/forum?id=WmF-fagWdD
9
SCALE: Causal Learning and Discovery of Robot Manipulation Skills using Simulation We propose SCALE, an approach for discovering and learning a diverse set of interpretable robot skills from a limited dataset. Rather than learning a single skill which may fail to capture all the modes in the data, we first identify the...
[ -0.048073090612888336, -0.02520679496228695, -0.00464152405038476, 0.02606126293540001, -0.011507449671626091, 0.01547329407185316, 0.016467077657580376, 0.022290460765361786, 0.013522879220545292, 0.032766975462436676, -0.013885099440813065, 0.002544827526435256, 0.0022406557109206915, 0....
corl_2023_WuBv9-IGDUA
WuBv9-IGDUA
corl
2,023
Multi-Resolution Sensing for Real-Time Control with Vision-Language Models
Leveraging sensing modalities across diverse spatial and temporal resolutions can improve performance of robotic manipulation tasks. Multi-spatial resolution sensing provides hierarchical information captured at different spatial scales and enables both coarse and precise motions. Simultaneously multi-temporal resoluti...
Saumya Saxena;Mohit Sharma;Oliver Kroemer
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Manipulation;Learning for manipulation
https://openreview.net/forum?id=WuBv9-IGDUA
19
Multi-Resolution Sensing for Real-Time Control with Vision-Language Models Leveraging sensing modalities across diverse spatial and temporal resolutions can improve performance of robotic manipulation tasks. Multi-spatial resolution sensing provides hierarchical information captured at different spatial scales and enab...
[ -0.02953764609992504, 0.0012720376253128052, 0.007852758280932903, 0.02173490636050701, -0.009721595793962479, 0.018697476014494896, -0.011622264049947262, 0.023590102791786194, 0.01834280602633953, 0.0007025194354355335, -0.040341440588235855, -0.013795753940939903, 0.023317279294133186, ...
corl_2023_X0cmlTh1Vl
X0cmlTh1Vl
corl
2,023
Waypoint-Based Imitation Learning for Robotic Manipulation
While imitation learning methods have seen a resurgent interest for robotic manipulation, the well-known problem of compounding errors continues to afflict behavioral cloning (BC). Waypoints can help address this problem by reducing the horizon of the learning problem for BC, and thus, the errors compounded over time. ...
Lucy Xiaoyang Shi;Archit Sharma;Tony Z. Zhao;Chelsea Finn
University of Southern California;Stanford University;Stanford University;Google
Poster
main
imitation learning;waypoints;long-horizon
https://github.com/lucys0/awe
https://openreview.net/forum?id=X0cmlTh1Vl
59
Waypoint-Based Imitation Learning for Robotic Manipulation While imitation learning methods have seen a resurgent interest for robotic manipulation, the well-known problem of compounding errors continues to afflict behavioral cloning (BC). Waypoints can help address this problem by reducing the horizon of the learning ...
[ -0.011981111019849777, -0.02375725843012333, -0.016909576952457428, -0.002212685765698552, 0.0012146480148658156, -0.03501167520880699, 0.027241656556725502, 0.04527853801846504, -0.009181481786072254, 0.03208627179265022, -0.02351502701640129, -0.04069478437304497, 0.029403099790215492, 0...
corl_2023_X7okQlJz9M
X7okQlJz9M
corl
2,023
Seeing-Eye Quadruped Navigation with Force Responsive Locomotion Control
Seeing-eye robots are very useful tools for guiding visually impaired people, potentially producing a huge societal impact given the low availability and high cost of real guide dogs. Although a few seeing-eye robot systems have already been demonstrated, none considered external tugs from humans, which frequently occu...
David DeFazio;Eisuke Hirota;Shiqi Zhang
State University of New York at Binghamton;New York University;State University of New York at Binghamton
Poster
main
seeing-eye robot;robotic guide dog;human-robot interaction;quadruped locomotion
https://github.com/bu-air-lab/guide_dog
https://openreview.net/forum?id=X7okQlJz9M
10
Seeing-Eye Quadruped Navigation with Force Responsive Locomotion Control Seeing-eye robots are very useful tools for guiding visually impaired people, potentially producing a huge societal impact given the low availability and high cost of real guide dogs. Although a few seeing-eye robot systems have already been demon...
[ -0.03532328084111214, 0.008969699963927269, 0.01870563067495823, 0.020017802715301514, -0.0121543500572443, -0.040188852697610855, -0.004539922811090946, 0.028906090185046196, 0.04175962507724762, 0.026128500699996948, -0.010842178016901016, -0.0008458479424007237, 0.027986612170934677, 0....
corl_2023_XEw-cnNsr6
XEw-cnNsr6
corl
2,023
DATT: Deep Adaptive Trajectory Tracking for Quadrotor Control
Precise arbitrary trajectory tracking for quadrotors is challenging due to unknown nonlinear dynamics, trajectory infeasibility, and actuation limits. To tackle these challenges, we present DATT, a learning-based approach that can precisely track arbitrary, potentially infeasible trajectories in the presence of large d...
Kevin Huang;Rwik Rana;Alexander Spitzer;Guanya Shi;Byron Boots
University of Washington;;University of Washington;University of Washington;
Oral
main
Quadrotor;Reinforcement Learning;Adaptive Control
https://github.com/KevinHuang8/DATT
https://openreview.net/forum?id=XEw-cnNsr6
27
DATT: Deep Adaptive Trajectory Tracking for Quadrotor Control Precise arbitrary trajectory tracking for quadrotors is challenging due to unknown nonlinear dynamics, trajectory infeasibility, and actuation limits. To tackle these challenges, we present DATT, a learning-based approach that can precisely track arbitrary, ...
[ -0.03477178514003754, -0.037805940955877304, -0.034641485661268234, 0.003448326140642166, -0.04925382509827614, -0.04258985444903374, -0.0015415088273584843, 0.009377026930451393, 0.007431817706674337, 0.04121238365769386, -0.039909373968839645, -0.017665108665823936, 0.01074984297156334, ...
corl_2023_XMQgwiJ7KSX
XMQgwiJ7KSX
corl
2,023
RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control
We study how vision-language models trained on Internet-scale data can be incorporated directly into end-to-end robotic control to boost generalization and enable emergent semantic reasoning. Our goal is to enable a single end-to-end trained model to both learn to map robot observations to actions and enjoy the benefit...
Brianna Zitkovich;Tianhe Yu;Sichun Xu;Peng Xu;Ted Xiao;Fei Xia;Jialin Wu;Paul Wohlhart;Stefan Welker;Ayzaan Wahid;Quan Vuong;Vincent Vanhoucke;Huong Tran;Radu Soricut;Anikait Singh;Jaspiar Singh;Pierre Sermanet;Pannag R Sanketi;Grecia Salazar;Michael S Ryoo;Krista Reymann;Kanishka Rao;Karl Pertsch;Igor Mordatch;Henryk ...
;Google Brain;;Google;;Google;Google;Graz University of Technology;;Robotics at Google;;Google;;Google;University of California, Berkeley;;Google;Google;;Google DeepMind;;;University of Southern California;;Google DeepMind;Google;Google;;;;;Google;Google;Research, Google;Google DeepMind;Google;;Google;;Research, Google...
Poster
main
vision-language models;robot manipulation;generalization
https://openreview.net/forum?id=XMQgwiJ7KSX
1,068
RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control We study how vision-language models trained on Internet-scale data can be incorporated directly into end-to-end robotic control to boost generalization and enable emergent semantic reasoning. Our goal is to enable a single end-to-end trained ...
[ -0.010120922699570656, 0.009824338369071484, 0.008851173333823681, 0.022540370002388954, -0.013661392033100128, 0.003994612954556942, -0.005157777573913336, 0.04163295030593872, 0.018017467111349106, 0.0022035250440239906, 0.004179977811872959, 0.0027387661393731833, -0.019370630383491516, ...
corl_2023_XsWGVbPfB4Z
XsWGVbPfB4Z
corl
2,023
4D-Former: Multimodal 4D Panoptic Segmentation
4D panoptic segmentation is a challenging but practically useful task that requires every point in a LiDAR point-cloud sequence to be assigned a semantic class label, and individual objects to be segmented and tracked over time. Existing approaches utilize only LiDAR inputs which convey limited information in regions w...
Ali Athar;Enxu Li;Sergio Casas;Raquel Urtasun
Waabi Innovation;Waabi;Department of Computer Science, University of Toronto;University of Toronto
Poster
main
Panoptic Segmentation;Sensor Fusion;Temporal Reasoning;Autonomous Driving
https://openreview.net/forum?id=XsWGVbPfB4Z
13
4D-Former: Multimodal 4D Panoptic Segmentation 4D panoptic segmentation is a challenging but practically useful task that requires every point in a LiDAR point-cloud sequence to be assigned a semantic class label, and individual objects to be segmented and tracked over time. Existing approaches utilize only LiDAR input...
[ -0.041429612785577774, 0.00906272791326046, -0.022100655362010002, 0.020787745714187622, -0.011889131739735603, -0.014195842668414116, 0.03760029003024101, -0.0037085155490785837, 0.002915299031883478, 0.02366885542869568, -0.020076585933566093, -0.012117067351937294, 0.03696206957101822, ...
corl_2023_ZFjgfJb_5c
ZFjgfJb_5c
corl
2,023
Embodied Lifelong Learning for Task and Motion Planning
A robot deployed in a home over long stretches of time faces a true lifelong learning problem. As it seeks to provide assistance to its users, the robot should leverage any accumulated experience to improve its own knowledge and proficiency. We formalize this setting with a novel formulation of lifelong learning for ta...
Jorge Mendez-Mendez;Leslie Pack Kaelbling;Tomás Lozano-Pérez
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
task and motion planning;lifelong learning;generative models
https://openreview.net/forum?id=ZFjgfJb_5c
20
Embodied Lifelong Learning for Task and Motion Planning A robot deployed in a home over long stretches of time faces a true lifelong learning problem. As it seeks to provide assistance to its users, the robot should leverage any accumulated experience to improve its own knowledge and proficiency. We formalize this sett...
[ -0.03739340603351593, -0.03295595571398735, -0.012801764532923698, 0.02231721580028534, -0.027218835428357124, -0.03301165625452995, 0.017369180917739868, 0.0006318492232821882, -0.005026943515986204, 0.04916471615433693, 0.003286312334239483, 0.0068279169499874115, -0.03624226897954941, 0...
corl_2023__A15qsPswaK
_A15qsPswaK
corl
2,023
HYDRA: Hybrid Robot Actions for Imitation Learning
Imitation Learning (IL) is a sample efficient paradigm for robot learning using expert demonstrations. However, policies learned through IL suffer from state distribution shift at test time, due to compounding errors in action prediction which lead to previously unseen states. Choosing an action representation for the ...
Suneel Belkhale;Yuchen Cui;Dorsa Sadigh
Stanford University;Stanford University;Stanford University
Poster
main
Imitation Learning;Robotics;Manipulation
https://sites.google.com/corp/view/hydra-il-2023
https://openreview.net/forum?id=_A15qsPswaK
40
HYDRA: Hybrid Robot Actions for Imitation Learning Imitation Learning (IL) is a sample efficient paradigm for robot learning using expert demonstrations. However, policies learned through IL suffer from state distribution shift at test time, due to compounding errors in action prediction which lead to previously unseen...
[ -0.0539175346493721, -0.03239117190241814, 0.022745883092284203, 0.019105803221464157, -0.04911337047815323, -0.009996362961828709, 0.0013326944317668676, 0.044604841619729996, 0.0013211459154263139, 0.002917145611718297, -0.05517401173710823, -0.041611477732658386, -0.016786867752671242, ...
corl_2023__DYsYC9smK
_DYsYC9smK
corl
2,023
DYNAMO-GRASP: DYNAMics-aware Optimization for GRASP Point Detection in Suction Grippers
In this research, we introduce a novel approach to the challenge of suction grasp point detection. Our method, exploiting the strengths of physics-based simulation and data-driven modeling, accounts for object dynamics during the grasping process, markedly enhancing the robot's capability to handle previously unseen ob...
Boling Yang;Soofiyan Atar;Markus Grotz;Byron Boots;Joshua Smith
Department of Computer Science, University of Washington;;University of Washington;;University of Washington
Poster
main
Suction Grasping;Manipulation;Deep Learning;Vision
https://github.com/dynamo-grasp/dynamo-grasp
https://openreview.net/forum?id=_DYsYC9smK
7
DYNAMO-GRASP: DYNAMics-aware Optimization for GRASP Point Detection in Suction Grippers In this research, we introduce a novel approach to the challenge of suction grasp point detection. Our method, exploiting the strengths of physics-based simulation and data-driven modeling, accounts for object dynamics during the gr...
[ -0.05253642797470093, -0.03385639190673828, -0.023720085620880127, 0.005410090554505587, -0.04680313542485237, -0.04313083365559578, -0.0037191445007920265, 0.022464755922555923, 0.021321846172213554, 0.04215654730796814, -0.026923982426524162, 0.035598862916231155, -0.0013607196742668748, ...
corl_2023__gZLyRGGuo
_gZLyRGGuo
corl
2,023
Learning Efficient Abstract Planning Models that Choose What to Predict
An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recent work has shown how to enable such bilevel planning by learning abstrac...
Nishanth Kumar;Willie McClinton;Rohan Chitnis;Tom Silver;Tomás Lozano-Pérez;Leslie Pack Kaelbling
The AI Institute;Massachusetts Institute of Technology;Meta;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Learning for TAMP;Abstraction Learning;Long-horizon Problems
https://github.com/Learning-and-Intelligent-Systems/predicators_behavior/releases/tag/corl-23-submission
https://openreview.net/forum?id=_gZLyRGGuo
25
Learning Efficient Abstract Planning Models that Choose What to Predict An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recen...
[ -0.0715978667140007, -0.022503770887851715, -0.011196412146091461, 0.0076692188158631325, -0.009902029298245907, -0.004481800831854343, -0.0007916538161225617, 0.01497785933315754, 0.014478596858680248, 0.025055553764104843, -0.0325629748404026, -0.03572496771812439, 0.005524241365492344, ...
corl_2023__xFJuqBId8c
_xFJuqBId8c
corl
2,023
Shelving, Stacking, Hanging: Relational Pose Diffusion for Multi-modal Rearrangement
We propose a system for rearranging objects in a scene to achieve a desired object-scene placing relationship, such as a book inserted in an open slot of a bookshelf. The pipeline generalizes to novel geometries, poses, and layouts of both scenes and objects, and is trained from demonstrations to operate directly on 3D...
Anthony Simeonov;Ankit Goyal;Lucas Manuelli;Yen-Chen Lin;Alina Sarmiento;Alberto Rodriguez Garcia;Pulkit Agrawal;Dieter Fox
Massachusetts Institute of Technology;NVIDIA;Boston Dynamics;Massachusetts Institute of Technology;;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Department of Computer Science
Poster
main
Object Rearrangement;Multi-modality;Manipulation;Point Clouds;Relations;Diffusion
https://github.com/anthonysimeonov/rpdiff
https://openreview.net/forum?id=_xFJuqBId8c
46
Shelving, Stacking, Hanging: Relational Pose Diffusion for Multi-modal Rearrangement We propose a system for rearranging objects in a scene to achieve a desired object-scene placing relationship, such as a book inserted in an open slot of a bookshelf. The pipeline generalizes to novel geometries, poses, and layouts of ...
[ -0.0007210990297608078, -0.014299974776804447, -0.022504273802042007, 0.01766088232398033, -0.0038811585400253534, 0.039189111441373825, -0.003301056567579508, 0.020681096240878105, 0.03244888037443161, -0.0173017717897892, -0.030699364840984344, -0.014539381489157677, -0.01504581980407238, ...
corl_2023_a0mFRgadGO
a0mFRgadGO
corl
2,023
Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidance
We propose BOSS, an approach that automatically learns to solve new long-horizon, complex, and meaningful tasks by growing a learned skill library with minimal supervision. Prior work in reinforcement learning require expert supervision, in the form of demonstrations or rich reward functions, to learn long-horizon task...
Jesse Zhang;Jiahui Zhang;Karl Pertsch;Ziyi Liu;Xiang Ren;Minsuk Chang;Shao-Hua Sun;Joseph J Lim
Amazon;University of Southern California;University of Southern California;University of Southern California;University of Southern California;Research, Google;National Taiwan University;Korea Advanced Institute of Science & Technology
Oral
main
Reinforcement Learning;Skill Learning;Large Language Models
https://openreview.net/forum?id=a0mFRgadGO
80
Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidance We propose BOSS, an approach that automatically learns to solve new long-horizon, complex, and meaningful tasks by growing a learned skill library with minimal supervision. Prior work in reinforcement learning require expert super...
[ -0.04149210825562477, -0.00851072371006012, -0.005372743587940931, -0.021509597077965736, -0.0015817934181541204, -0.011229686439037323, 0.02691027522087097, -0.005479826126247644, -0.0028586348053067923, -0.006643765140324831, -0.03892212733626366, -0.03298138454556465, -0.02758070454001426...
corl_2023_afF8RGcBBP
afF8RGcBBP
corl
2,023
PlayFusion: Skill Acquisition via Diffusion from Language-Annotated Play
Learning from unstructured and uncurated data has become the dominant paradigm for generative approaches in language or vision. Such unstructured and unguided behavior data, commonly known as play, is also easier to collect in robotics but much more difficult to learn from due to its inherently multimodal, noisy, and s...
Lili Chen;Shikhar Bahl;Deepak Pathak
Carnegie Mellon University;Meta Facebook;Carnegie Mellon University
Poster
main
Diffusion Models;Learning from Play;Language-Driven Robotics
https://openreview.net/forum?id=afF8RGcBBP
53
PlayFusion: Skill Acquisition via Diffusion from Language-Annotated Play Learning from unstructured and uncurated data has become the dominant paradigm for generative approaches in language or vision. Such unstructured and unguided behavior data, commonly known as play, is also easier to collect in robotics but much mo...
[ -0.05171390622854233, -0.0016519720666110516, -0.03271622955799103, 0.015101178549230099, -0.0438849963247776, 0.007492231670767069, -0.004224200267344713, 0.021547460928559303, -0.00039419508539140224, 0.019662057980895042, -0.023468775674700737, -0.03494279831647873, -0.04463915526866913, ...
corl_2023_b-cto-fetlz
b-cto-fetlz
corl
2,023
HomeRobot: Open-Vocabulary Mobile Manipulation
HomeRobot (noun): An affordable compliant robot that navigates homes and manipulates a wide range of objects in order to complete everyday tasks. Open-Vocabulary Mobile Manipulation (OVMM) is the problem of picking any object in any unseen environment, and placing it in a commanded location. This is a foundational cha...
Sriram Yenamandra;Arun Ramachandran;Karmesh Yadav;Austin S Wang;Mukul Khanna;Theophile Gervet;Tsung-Yen Yang;Vidhi Jain;Alexander Clegg;John M Turner;Zsolt Kira;Manolis Savva;Angel X Chang;Devendra Singh Chaplot;Dhruv Batra;Roozbeh Mottaghi;Yonatan Bisk;Chris Paxton
Georgia Institute of Technology;Georgia Institute of Technology;Meta AI;Meta Facebook;Georgia Institute of Technology;Carnegie Mellon University;Meta AI;Google;Meta AI;;Georgia Tech Research Institute;Simon Fraser University;Simon Fraser University;Georgia Institute of Technology;University of Washington;Meta;Meta Plat...
Poster
main
benchmark;mobile manipulation;sim2real
https://github.com/facebookresearch/home-robot
https://openreview.net/forum?id=b-cto-fetlz
98
HomeRobot: Open-Vocabulary Mobile Manipulation HomeRobot (noun): An affordable compliant robot that navigates homes and manipulates a wide range of objects in order to complete everyday tasks. Open-Vocabulary Mobile Manipulation (OVMM) is the problem of picking any object in any unseen environment, and placing it in a...
[ -0.03486687317490578, -0.023288797587156296, -0.02688918076455593, 0.008271407335996628, -0.005732189863920212, 0.01767977885901928, -0.00984420720487833, -0.043735187500715256, 0.01531110517680645, 0.02715447172522545, -0.026756536215543747, 0.024008873850107193, 0.010232669301331043, -0....
corl_2023_b1tl3aOt2R2
b1tl3aOt2R2
corl
2,023
GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields
It is a long-standing problem in robotics to develop agents capable of executing diverse manipulation tasks from visual observations in unstructured real-world environments. To achieve this goal, the robot will need to have a comprehensive understanding of the 3D structure and semantics of the scene. In this work, we ...
Yanjie Ze;Ge Yan;Yueh-Hua Wu;Annabella Macaluso;Yuying Ge;Jianglong Ye;Nicklas Hansen;Li Erran Li;Xiaolong Wang
Shanghai Jiaotong University;University of California, San Diego;;University of California, San Diego;University of Hong Kong;University of California, San Diego;University of California, San Diego;Columbia University;University of California, San Diego
Oral
main
Robotic Manipulation;Neural Radiance Field;Behavior Cloning
https://github.com/YanjieZe/GNFactor
https://openreview.net/forum?id=b1tl3aOt2R2
88
GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields It is a long-standing problem in robotics to develop agents capable of executing diverse manipulation tasks from visual observations in unstructured real-world environments. To achieve this goal, the robot will need to have a comprehensi...
[ -0.04139939323067665, -0.007894028909504414, 0.012201926670968533, -0.013972295448184013, 0.004834468476474285, -0.007966659963130951, -0.02071785368025303, 0.017440402880311012, -0.0025806529447436333, 0.027309073135256767, -0.03943837061524391, -0.038712065666913986, -0.04147202521562576, ...
corl_2023_bIvIUNH9VQ
bIvIUNH9VQ
corl
2,023
Hijacking Robot Teams Through Adversarial Communication
Communication is often necessary for robot teams to collaborate and complete a decentralized task. Multi-agent reinforcement learning (MARL) systems allow agents to learn how to collaborate and communicate to complete a task. These domains are ubiquitous and include safety-critical domains such as wildfire fighting, tr...
Zixuan Wu;Sean Charles Ye;Byeolyi Han;Matthew Gombolay
Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology
Oral
main
Adversarial Attacks;Multi-Agent Reinforcement Learning
https://openreview.net/forum?id=bIvIUNH9VQ
0
Hijacking Robot Teams Through Adversarial Communication Communication is often necessary for robot teams to collaborate and complete a decentralized task. Multi-agent reinforcement learning (MARL) systems allow agents to learn how to collaborate and communicate to complete a task. These domains are ubiquitous and inclu...
[ -0.08085666596889496, -0.020639533177018166, 0.01738455332815647, -0.006163193844258785, -0.018235286697745323, -0.012945943512022495, 0.026576172560453415, 0.004808493424206972, -0.03798709809780121, -0.001525771920569241, -0.044127173721790314, -0.03643358498811722, -0.033363547176122665, ...
corl_2023_cjEI5qXoT0
cjEI5qXoT0
corl
2,023
Context-Aware Entity Grounding with Open-Vocabulary 3D Scene Graphs
We present an Open-Vocabulary 3D Scene Graph (OVSG), a formal framework for grounding a variety of entities, such as object instances, agents, and regions, with free-form text-based queries. Unlike conventional semantic-based object localization approaches, our system facilitates context-aware entity localization, allo...
Haonan Chang;Kowndinya Boyalakuntla;Shiyang Lu;Siwei Cai;Eric Pu Jing;Shreesh Keskar;Shijie Geng;Adeeb Abbas;Lifeng Zhou;Kostas Bekris;Abdeslam Boularias
Rutgers, New Brunswick;Rutgers University;Rutgers University - New Brunswick;Drexel University;Rutgers University;Rutgers University;ByteDance Inc.;Northeastern University;;Rutgers University;, Rutgers University
Poster
main
Open-Vocabulary Semantic;Scene Graph;Object Grounding
https://github.com/changhaonan/OVSG
https://openreview.net/forum?id=cjEI5qXoT0
29
Context-Aware Entity Grounding with Open-Vocabulary 3D Scene Graphs We present an Open-Vocabulary 3D Scene Graph (OVSG), a formal framework for grounding a variety of entities, such as object instances, agents, and regions, with free-form text-based queries. Unlike conventional semantic-based object localization approa...
[ -0.04108179360628128, -0.03295101970434189, -0.023573655635118484, 0.002742042066529393, -0.0025815663393586874, -0.07572595030069351, 0.01633596606552601, -0.012270580977201462, 0.03523954376578331, -0.006656255107372999, 0.01492191944271326, -0.0014663762412965298, 0.0497521348297596, -0...
corl_2023_ckeT8cMz_A
ckeT8cMz_A
corl
2,023
REBOOT: Reuse Data for Bootstrapping Efficient Real-World Dexterous Manipulation
Dexterous manipulation tasks involving contact-rich interactions pose a significant challenge for both model-based control systems and imitation learning algorithms. The complexity arises from the need for multi-fingered robotic hands to dynamically establish and break contacts, balance forces on the non-prehensile obj...
Zheyuan Hu;Aaron Rovinsky;Jianlan Luo;Vikash Kumar;Abhishek Gupta;Sergey Levine
University of California, Berkeley;University of California, Berkeley;Google;Meta Facebook;University of Washington;Google
Poster
main
Dexterous Manipulation;Reinforcement Learning;Sample-Efficient RL
https://openreview.net/forum?id=ckeT8cMz_A
11
REBOOT: Reuse Data for Bootstrapping Efficient Real-World Dexterous Manipulation Dexterous manipulation tasks involving contact-rich interactions pose a significant challenge for both model-based control systems and imitation learning algorithms. The complexity arises from the need for multi-fingered robotic hands to d...
[ -0.03190680593252182, 0.000602930027525872, -0.031687766313552856, 0.013790384866297245, -0.029862435534596443, 0.005749796517193317, 0.044720638543367386, 0.034900352358818054, 0.012786452658474445, -0.0009041098528541625, -0.03725503012537956, 0.006292833015322685, 0.005124620161950588, ...
corl_2023_dIgCPoy8E3
dIgCPoy8E3
corl
2,023
Cross-Dataset Sensor Alignment: Making Visual 3D Object Detector Generalizable
While camera-based 3D object detection has evolved rapidly, these models are susceptible to overfitting to specific sensor setups. For example, in autonomous driving, most datasets are collected using a single sensor configuration. This paper evaluates the generalization capability of camera-based 3D object detectors, ...
Liangtao Zheng;Yicheng Liu;Yue Wang;Hang Zhao
Wuhan University;Tsinghua University;NVIDIA;Tsinghua University
Poster
main
3D object detection;Model Generalization;Autonomous Driving
https://openreview.net/forum?id=dIgCPoy8E3
3
Cross-Dataset Sensor Alignment: Making Visual 3D Object Detector Generalizable While camera-based 3D object detection has evolved rapidly, these models are susceptible to overfitting to specific sensor setups. For example, in autonomous driving, most datasets are collected using a single sensor configuration. This pape...
[ -0.04672098904848099, -0.04150545597076416, -0.0626593753695488, 0.014844218268990517, -0.02963372692465782, -0.020570365712046623, -0.02631475031375885, 0.0027513776440173388, -0.009692508727312088, -0.0004408016975503415, -0.04117720574140549, -0.04321965202689171, 0.007444904185831547, ...
corl_2023_dgwvY3H8PAS
dgwvY3H8PAS
corl
2,023
Dynamic Handover: Throw and Catch with Bimanual Hands
Humans throw and catch objects all the time. However, such a seemingly common skill introduces a lot of challenges for robots to achieve: The robots need to operate such dynamic actions at high-speed, collaborate precisely, and interact with diverse objects. In this paper, we design a system with two multi-finger hands...
Binghao Huang;Yuanpei Chen;Tianyu Wang;Yuzhe Qin;Yaodong Yang;Nikolay Atanasov;Xiaolong Wang
University of California, San Diego;South China University of Technology;University of California, San Diego;University of California, San Diego;Peking University;University of California, San Diego;University of California, San Diego
Poster
main
Bimanual Dexterous Manipulation;Sim-to-Real Transfer
https://openreview.net/forum?id=dgwvY3H8PAS
50
Dynamic Handover: Throw and Catch with Bimanual Hands Humans throw and catch objects all the time. However, such a seemingly common skill introduces a lot of challenges for robots to achieve: The robots need to operate such dynamic actions at high-speed, collaborate precisely, and interact with diverse objects. In this...
[ -0.06231708452105522, -0.03884604945778847, -0.002525364514440298, 0.008007819764316082, -0.019423024728894234, -0.01390807330608368, -0.005839905701577663, 0.03186415880918503, -0.004103717394173145, 0.018726691603660583, -0.0347423292696476, 0.006174144800752401, 0.0008721560006961226, 0...
corl_2023_diOr96f65N
diOr96f65N
corl
2,023
Quantifying Assistive Robustness Via the Natural-Adversarial Frontier
Our ultimate goal is to build robust policies for robots that assist people. What makes this hard is that people can behave unexpectedly at test time, potentially interacting with the robot outside its training distribution and leading to failures. Even just measuring robustness is a challenge. Adversarial perturbation...
Jerry Zhi-Yang He;Daniel S. Brown;Zackory Erickson;Anca Dragan
;University of Utah;Carnegie Mellon University;University of California, Berkeley
Poster
main
assistive robots;safety;human-robot interaction;adversarial robustness
https://openreview.net/forum?id=diOr96f65N
0
Quantifying Assistive Robustness Via the Natural-Adversarial Frontier Our ultimate goal is to build robust policies for robots that assist people. What makes this hard is that people can behave unexpectedly at test time, potentially interacting with the robot outside its training distribution and leading to failures. E...
[ -0.018746288493275642, 0.0037711886689066887, -0.023094359785318375, -0.032477039843797684, -0.036176715046167374, -0.0010256345849484205, 0.008715212345123291, -0.0031728523317724466, 0.039781033992767334, 0.04721852391958237, -0.05530441179871559, -0.014817768707871437, -0.0354901775717735...
corl_2023_dk-2R1f_LR
dk-2R1f_LR
corl
2,023
MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrations
Imitation learning from a large set of human demonstrations has proved to be an effective paradigm for building capable robot agents. However, the demonstrations can be extremely costly and time-consuming to collect. We introduce MimicGen, a system for automatically synthesizing large-scale, rich datasets from only a s...
Ajay Mandlekar;Soroush Nasiriany;Bowen Wen;Iretiayo Akinola;Yashraj Narang;Linxi Fan;Yuke Zhu;Dieter Fox
NVIDIA;University of Texas, Austin;NVIDIA;NVIDIA;NVIDIA;NVIDIA;Computer Science Department, University of Texas, Austin;Department of Computer Science
Poster
main
Imitation Learning;Manipulation
https://github.com/NVlabs/mimicgen_environments
https://openreview.net/forum?id=dk-2R1f_LR
120
MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrations Imitation learning from a large set of human demonstrations has proved to be an effective paradigm for building capable robot agents. However, the demonstrations can be extremely costly and time-consuming to collect. We introduce ...
[ -0.01899261772632599, -0.03451525419950485, -0.02707693912088871, -0.012006509117782116, -0.014701282605528831, -0.03403536230325699, -0.019121820107102394, -0.004900427535176277, 0.023662328720092773, 0.008001263253390789, -0.024788226932287216, -0.021447446197271347, 0.0018688071286305785,...
corl_2023_dxOaNO8bge
dxOaNO8bge
corl
2,023
A Data-Efficient Visual-Audio Representation with Intuitive Fine-tuning for Voice-Controlled Robots
A command-following robot that serves people in everyday life must continually improve itself in deployment domains with minimal help from its end users, instead of engineers. Previous methods are either difficult to continuously improve after the deployment or require a large number of new labels during fine-tuning. M...
Peixin Chang;Shuijing Liu;Tianchen Ji;Neeloy Chakraborty;Kaiwen Hong;Katherine Rose Driggs-Campbell
University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;UIUC;
Poster
main
Command Following;Multimodal Representation;Reinforcement Learning;Human-in-the-Loop
https://openreview.net/forum?id=dxOaNO8bge
8
A Data-Efficient Visual-Audio Representation with Intuitive Fine-tuning for Voice-Controlled Robots A command-following robot that serves people in everyday life must continually improve itself in deployment domains with minimal help from its end users, instead of engineers. Previous methods are either difficult to con...
[ -0.07934225350618362, -0.017692947760224342, -0.015802955254912376, 0.01727190986275673, -0.012696631252765656, -0.029416514560580254, 0.017898788675665855, 0.01319252047687769, 0.023035451769828796, 0.03751915320754051, -0.010572729632258415, -0.014857959002256393, -0.01000198908150196, 0...
corl_2023_eE3fsO5Mi2
eE3fsO5Mi2
corl
2,023
Stealthy Terrain-Aware Multi-Agent Active Search
Stealthy multi-agent active search is the problem of making efficient sequential data-collection decisions to identify an unknown number of sparsely located targets while adapting to new sensing information and concealing the search agents' location from the targets. This problem is applicable to reconnaissance tasks w...
Nikhil Angad Bakshi;Jeff Schneider
Carnegie Mellon University;Carnegie Mellon University
Poster
main
Reconnaissance;Adversarial Search;Multi-robot;Active Learning
https://github.com/bakshienator77/Stealthy-Terrain-Aware-Reconnaissance-and-Search
https://openreview.net/forum?id=eE3fsO5Mi2
1
Stealthy Terrain-Aware Multi-Agent Active Search Stealthy multi-agent active search is the problem of making efficient sequential data-collection decisions to identify an unknown number of sparsely located targets while adapting to new sensing information and concealing the search agents' location from the targets. Thi...
[ 0.004129193257540464, -0.009943582117557526, -0.038284651935100555, 0.010856008157134056, 0.006461466196924448, -0.003640393726527691, 0.024691365659236908, 0.00080011849058792, 0.018416110426187515, 0.01820196956396103, -0.02694449946284294, -0.015352965332567692, -0.02856452204287052, 0....
corl_2023_efaE7iJ2GJv
efaE7iJ2GJv
corl
2,023
PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation
The ability for robots to comprehend and execute manipulation tasks based on natural language instructions is a long-term goal in robotics. The dominant approaches for language-guided manipulation use 2D image representations, which face difficulties in combining multi-view cameras and inferring precise 3D positions an...
Shizhe Chen;Ricardo Garcia Pinel;Cordelia Schmid;Ivan Laptev
INRIA;INRIA;Inria;INRIA Paris
Poster
main
Robotic manipulation;3D point clouds;language-guided policy
https://openreview.net/forum?id=efaE7iJ2GJv
37
PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation The ability for robots to comprehend and execute manipulation tasks based on natural language instructions is a long-term goal in robotics. The dominant approaches for language-guided manipulation use 2D image representations, which face difficulties in...
[ -0.017811501398682594, 0.014160925522446632, -0.03685518726706505, -0.031779322773218155, -0.036579325795173645, 0.021425295621156693, 0.0020896559581160545, -0.00855172798037529, 0.00008081900159595534, -0.004328737501055002, -0.018059777095913887, -0.01525517925620079, 0.027917252853512764...
corl_2023_eyykI3UIHa
eyykI3UIHa
corl
2,023
NOIR: Neural Signal Operated Intelligent Robots for Everyday Activities
We present Neural Signal Operated Intelligent Robots (NOIR), a general-purpose, intelligent brain-robot interface system that enables humans to command robots to perform everyday activities through brain signals. Through this interface, humans communicate their intended objects of interest and actions to the robots usi...
Ruohan Zhang;Sharon Lee;Minjune Hwang;Ayano Hiranaka;Chen Wang;Wensi Ai;Jin Jie Ryan Tan;Shreya Gupta;Yilun Hao;Gabrael Levine;Ruohan Gao;Anthony Norcia;Li Fei-Fei;Jiajun Wu
Stanford University;;Stanford University;;Computer Science Department, Stanford University;Stanford University;Stanford University;Stanford University;Stanford University;Stanford University;Stanford University;Stanford University;Stanford University;Stanford University
Poster
main
Brain-Robot Interface;Human-Robot Interaction
https://openreview.net/forum?id=eyykI3UIHa
18
NOIR: Neural Signal Operated Intelligent Robots for Everyday Activities We present Neural Signal Operated Intelligent Robots (NOIR), a general-purpose, intelligent brain-robot interface system that enables humans to command robots to perform everyday activities through brain signals. Through this interface, humans comm...
[ -0.059163883328437805, -0.025988226756453514, 0.025685599073767662, 0.028144456446170807, -0.00511631416156888, -0.007149599492549896, -0.021089427173137665, -0.024437256157398224, 0.0061471424996852875, 0.05277085304260254, -0.004688378423452377, -0.0010048213880509138, -0.00532437115907669...
corl_2023_f55MlAT1Lu
f55MlAT1Lu
corl
2,023
BridgeData V2: A Dataset for Robot Learning at Scale
We introduce BridgeData V2, a large and diverse dataset of robotic manipulation behaviors designed to facilitate research in scalable robot learning. BridgeData V2 contains 53,896 trajectories collected across 24 environments on a publicly available low-cost robot. Unlike many existing robotic manipulation datasets, Br...
Homer Rich Walke;Kevin Black;Tony Z. Zhao;Quan Vuong;Chongyi Zheng;Philippe Hansen-Estruch;Andre Wang He;Vivek Myers;Moo Jin Kim;Max Du;Abraham Lee;Kuan Fang;Chelsea Finn;Sergey Levine
University of California, Berkeley;University of California, Berkeley;Stanford University;;Carnegie Mellon University;;UC Berkeley, University of California, Berkeley;University of California, Berkeley;Stanford University;Stanford University;University of California, Berkeley;;Google;Google
Poster
main
Datasets;Manipulation;Imitation Learning;Offline Reinforcement Learning
https://github.com/rail-berkeley/bridge_data_v2
https://openreview.net/forum?id=f55MlAT1Lu
168
BridgeData V2: A Dataset for Robot Learning at Scale We introduce BridgeData V2, a large and diverse dataset of robotic manipulation behaviors designed to facilitate research in scalable robot learning. BridgeData V2 contains 53,896 trajectories collected across 24 environments on a publicly available low-cost robot. U...
[ -0.07169246673583984, -0.01168044377118349, -0.014956208877265453, 0.017427071928977966, -0.00921426061540842, -0.000899080652743578, -0.019504843279719353, 0.00935932993888855, 0.014843896962702274, -0.012616376392543316, -0.041705172508955, 0.009087909013032913, 0.022518547251820564, 0.0...
corl_2023_fNLBmtyBiC
fNLBmtyBiC
corl
2,023
A Bayesian approach to breaking things: efficiently predicting and repairing failure modes via sampling
Before autonomous systems can be deployed in safety-critical applications, we must be able to understand and verify the safety of these systems. For cases where the risk or cost of real-world testing is prohibitive, we propose a simulation-based framework for a) predicting ways in which an autonomous system is likely t...
Charles Dawson;Chuchu Fan
Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Automatic design tools;root-cause failure analysis;optimization-as-inference
https://github.com/MIT-REALM/architect_corl_23
https://openreview.net/forum?id=fNLBmtyBiC
11
A Bayesian approach to breaking things: efficiently predicting and repairing failure modes via sampling Before autonomous systems can be deployed in safety-critical applications, we must be able to understand and verify the safety of these systems. For cases where the risk or cost of real-world testing is prohibitive, ...
[ -0.002399425720795989, -0.01436565537005663, -0.01608649268746376, 0.030632788315415382, 0.0019585208501666784, -0.013215262442827225, 0.07256034761667252, 0.041946571320295334, 0.007453788071870804, 0.08244802057743073, -0.05384981632232666, 0.005186277441680431, 0.021030331030488014, -0....
corl_2023_fSmkKmWM5Ry
fSmkKmWM5Ry
corl
2,023
Stochastic Occupancy Grid Map Prediction in Dynamic Scenes
This paper presents two variations of a novel stochastic prediction algorithm that enables mobile robots to accurately and robustly predict the future state of complex dynamic scenes. The proposed algorithm uses a variational autoencoder to predict a range of possible future states of the environment. The algorithm tak...
Zhanteng Xie;Philip Dames
Temple University;Temple University
Poster
main
Environment Prediction;Probabilistic Inference;Robot Learning
https://github.com/TempleRAIL/SOGMP
https://openreview.net/forum?id=fSmkKmWM5Ry
5
Stochastic Occupancy Grid Map Prediction in Dynamic Scenes This paper presents two variations of a novel stochastic prediction algorithm that enables mobile robots to accurately and robustly predict the future state of complex dynamic scenes. The proposed algorithm uses a variational autoencoder to predict a range of p...
[ -0.03900826722383499, -0.0242126677185297, -0.03863604739308357, 0.02547820471227169, -0.026017917320132256, -0.0474947951734066, -0.004306077025830746, 0.014153523370623589, 0.022426029667258263, 0.040497131645679474, -0.03346224129199982, 0.0048620752058923244, 0.0036523721646517515, -0....
corl_2023_fa7FzDjhzs9
fa7FzDjhzs9
corl
2,023
HACMan: Learning Hybrid Actor-Critic Maps for 6D Non-Prehensile Manipulation
Manipulating objects without grasping them is an essential component of human dexterity, referred to as non-prehensile manipulation. Non-prehensile manipulation may enable more complex interactions with the objects, but also presents challenges in reasoning about gripper-object interactions. In this work, we introduce ...
Wenxuan Zhou;Bowen Jiang;Fan Yang;Chris Paxton;David Held
Meta AI;Carnegie Mellon University;Carnegie Mellon University;Meta Platforms;Carnegie Mellon University
Oral
main
Action Representation;Reinforcement Learning with 3D Vision;Non-prehensile Manipulation
https://openreview.net/forum?id=fa7FzDjhzs9
22
HACMan: Learning Hybrid Actor-Critic Maps for 6D Non-Prehensile Manipulation Manipulating objects without grasping them is an essential component of human dexterity, referred to as non-prehensile manipulation. Non-prehensile manipulation may enable more complex interactions with the objects, but also presents challenge...
[ -0.03886471688747406, -0.009825657121837139, 0.0038910333532840014, -0.003970860969275236, 0.00017362539074383676, -0.013100876472890377, 0.017352275550365448, -0.0169691015034914, -0.012288914062082767, 0.027150563895702362, -0.042550478130578995, -0.029522588476538658, 0.01626661792397499,...
corl_2023_flyQ0v8cgC
flyQ0v8cgC
corl
2,023
Continual Vision-based Reinforcement Learning with Group Symmetries
Continual reinforcement learning aims to sequentially learn a variety of tasks, retaining the ability to perform previously encountered tasks while simultaneously developing new policies for novel tasks. However, current continual RL approaches overlook the fact that certain tasks are identical under basic group operat...
Shiqi Liu;Mengdi Xu;Peide Huang;Xilun Zhang;Yongkang Liu;Kentaro Oguchi;Ding Zhao
;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Toyota;Toyota Motor North America;Carnegie Mellon University
Oral
main
Continual Learning;Symmetry;Manipulation
https://openreview.net/forum?id=flyQ0v8cgC
10
Continual Vision-based Reinforcement Learning with Group Symmetries Continual reinforcement learning aims to sequentially learn a variety of tasks, retaining the ability to perform previously encountered tasks while simultaneously developing new policies for novel tasks. However, current continual RL approaches overloo...
[ -0.0655997171998024, 0.002145870588719845, -0.009587932378053665, 0.030042188242077827, -0.01567855291068554, -0.005195746198296547, 0.04138334095478058, 0.009323122911155224, 0.002209790050983429, 0.012117319740355015, -0.01985158585011959, -0.02763150818645954, -0.005930820945650339, 0.0...
corl_2023_fvXFBCHVGn
fvXFBCHVGn
corl
2,023
Dynamic Multi-Team Racing: Competitive Driving on 1/10-th Scale Vehicles via Learning in Simulation
Autonomous racing is a challenging task that requires vehicle handling at the dynamic limits of friction. While single-agent scenarios like Time Trials are solved competitively with classical model-based or model-free feedback control, multi-agent wheel-to-wheel racing poses several challenges including planning over u...
Peter Werner;Tim Seyde;Paul Drews;Thomas Matrai Balch;Igor Gilitschenski;Wilko Schwarting;Guy Rosman;Sertac Karaman;Daniela Rus
Computer Science and Artificial Intelligence Laboratory, Electrical Engineering & Computer Science;Massachusetts Institute of Technology;Toyota Research Institute;Toyota Research Institute;University of Toronto;;Toyota Research Institute;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Multi-Agent;Reinforcement Learning;Sim-to-Real Transfer;Autonomous Racing
https://openreview.net/forum?id=fvXFBCHVGn
6
Dynamic Multi-Team Racing: Competitive Driving on 1/10-th Scale Vehicles via Learning in Simulation Autonomous racing is a challenging task that requires vehicle handling at the dynamic limits of friction. While single-agent scenarios like Time Trials are solved competitively with classical model-based or model-free fe...
[ -0.05966377630829811, -0.02773052267730236, -0.004948528483510017, 0.020094309002161026, -0.03752438724040985, -0.04056386277079582, 0.013171060010790825, 0.02778680808842182, 0.0014353077858686447, 0.001504493411630392, -0.037599436938762665, 0.02054460160434246, -0.022627204656600952, 0....
corl_2023_fviZhMCr62
fviZhMCr62
corl
2,023
Tell Me Where to Go: A Composable Framework for Context-Aware Embodied Robot Navigation
Humans have the remarkable ability to navigate through unfamiliar environments by solely relying on our prior knowledge and descriptions of the environment. For robots to perform the same type of navigation, they need to be able to associate natural language descriptions with their associated physical environment with ...
Harel Biggie;Ajay Narasimha Mopidevi;Dusty Woods;Chris Heckman
University of Colorado at Boulder;University of Colorado at Boulder;;University of Colorado at Boulder
Poster
main
Natural language;navigation;contextual navigation
https://github.com/arpg/navcon
https://openreview.net/forum?id=fviZhMCr62
13
Tell Me Where to Go: A Composable Framework for Context-Aware Embodied Robot Navigation Humans have the remarkable ability to navigate through unfamiliar environments by solely relying on our prior knowledge and descriptions of the environment. For robots to perform the same type of navigation, they need to be able to ...
[ -0.06281169503927231, -0.027951577678322792, 0.0033469072077423334, -0.001874174689874053, -0.02063225954771042, -0.03463605418801308, -0.046567291021347046, -0.0368393175303936, 0.006119658704847097, -0.006413738708943129, -0.027186036109924316, -0.007580721750855446, -0.009321860037744045,...
corl_2023_gFXVysXh48K
gFXVysXh48K
corl
2,023
Efficient Sim-to-real Transfer of Contact-Rich Manipulation Skills with Online Admittance Residual Learning
Learning contact-rich manipulation skills is essential. Such skills require the robots to interact with the environment with feasible manipulation trajectories and suitable compliance control parameters to enable safe and stable contact. However, learning these skills is challenging due to data inefficiency in the real...
Xiang Zhang;Changhao Wang;Lingfeng Sun;Zheng Wu;Xinghao Zhu;Masayoshi Tomizuka
University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;;University of California, Berkeley
Poster
main
Contact-rich Manipulation;Compliance Control
https://openreview.net/forum?id=gFXVysXh48K
24
Efficient Sim-to-real Transfer of Contact-Rich Manipulation Skills with Online Admittance Residual Learning Learning contact-rich manipulation skills is essential. Such skills require the robots to interact with the environment with feasible manipulation trajectories and suitable compliance control parameters to enable...
[ -0.07322298735380173, 0.007137392181903124, 0.017945177853107452, 0.0002568467170931399, -0.0038922824896872044, -0.016900457441806793, 0.0251842699944973, 0.014237802475690842, 0.018111594021320343, 0.021171797066926956, -0.04541304334998131, 0.00972608383744955, 0.013109873048961163, 0.0...
corl_2023_gVBvtRqU1_
gVBvtRqU1_
corl
2,023
OVIR-3D: Open-Vocabulary 3D Instance Retrieval Without Training on 3D Data
This work presents OVIR-3D, a straightforward yet effective method for open-vocabulary 3D object instance retrieval without using any 3D data for training. Given a language query, the proposed method is able to return a ranked set of 3D object instance segments based on the feature similarity of the instance and the te...
Shiyang Lu;Haonan Chang;Eric Pu Jing;Abdeslam Boularias;Kostas Bekris
Rutgers University - New Brunswick;Rutgers, New Brunswick;Rutgers University;, Rutgers University;Rutgers University
Poster
main
Open-Vocabulary;3D Instance Retrieval
https://github.com/shiyoung77/OVIR-3D/
https://openreview.net/forum?id=gVBvtRqU1_
61
OVIR-3D: Open-Vocabulary 3D Instance Retrieval Without Training on 3D Data This work presents OVIR-3D, a straightforward yet effective method for open-vocabulary 3D object instance retrieval without using any 3D data for training. Given a language query, the proposed method is able to return a ranked set of 3D object i...
[ -0.00441081402823329, -0.03298521786928177, -0.013734708540141582, 0.02345128357410431, -0.00349988485686481, 0.03835490345954895, 0.0662991926074028, -0.01823684386909008, 0.03727731481194496, -0.008607023395597935, -0.020255042240023613, -0.02012719213962555, 0.01888522319495678, 0.01565...
corl_2023_g_PPHV_GkX
g_PPHV_GkX
corl
2,023
Hierarchical Planning for Rope Manipulation using Knot Theory and a Learned Inverse Model
This work considers planning the manipulation of deformable 1-dimensional objects, such as ropes or cables, specifically to tie knots. We propose TWISTED: Tying With Inverse model and Search in Topological space Excluding Demos, a hierarchical planning approach which, at the high level, uses ideas from knot-theory to p...
Matan Sudry;Tom Jurgenson;Aviv Tamar;Erez Karpas
Technion, Technion;Technion;Technion, Technion;Technion - Israel Institute of Technology, Technion
Poster
main
Manipulation;Robot Learning and Planning
https://openreview.net/forum?id=g_PPHV_GkX
4
Hierarchical Planning for Rope Manipulation using Knot Theory and a Learned Inverse Model This work considers planning the manipulation of deformable 1-dimensional objects, such as ropes or cables, specifically to tie knots. We propose TWISTED: Tying With Inverse model and Search in Topological space Excluding Demos, a...
[ -0.020533965900540352, 0.0032823376823216677, 0.008624641224741936, 0.040730081498622894, -0.002543283859267831, -0.02828582003712654, -0.005222647450864315, 0.027685193344950676, -0.009347272105515003, 0.05165399983525276, -0.03810233250260353, -0.01373936329036951, -0.005429113283753395, ...
corl_2023_gdkKi_F55h
gdkKi_F55h
corl
2,023
SA6D: Self-Adaptive Few-Shot 6D Pose Estimator for Novel and Occluded Objects
To enable meaningful robotic manipulation of objects in the real-world, 6D pose estimation is one of the critical aspects. Most existing approaches have difficulties to extend predictions to scenarios where novel object instances are continuously introduced, especially with heavy occlusions. In this work, we propose a ...
Ning Gao;Vien Anh Ngo;Hanna Ziesche;Gerhard Neumann
Robert Bosch GmbH, Bosch;Bosch Center for Artificial Intelligence;Karlsruhe Institute of Technology;Robert Bosch GmbH, Bosch
Poster
main
https://openreview.net/forum?id=gdkKi_F55h
6
SA6D: Self-Adaptive Few-Shot 6D Pose Estimator for Novel and Occluded Objects To enable meaningful robotic manipulation of objects in the real-world, 6D pose estimation is one of the critical aspects. Most existing approaches have difficulties to extend predictions to scenarios where novel object instances are continuo...
[ -0.0021061303559690714, -0.014371242374181747, -0.02083154395222664, 0.005536757409572601, -0.005352048669010401, -0.06231442466378212, 0.04724939912557602, -0.00036800967063754797, 0.034364838153123856, 0.03097700886428356, -0.018398793414235115, 0.002766126301139593, -0.019516056403517723,...
corl_2023_h-geaPzuJu
h-geaPzuJu
corl
2,023
DROID: Learning from Offline Heterogeneous Demonstrations via Reward-Policy Distillation
Offline Learning from Demonstrations (OLfD) is valuable in domains where trial-and-error learning is infeasible or specifying a cost function is difficult, such as robotic surgery, autonomous driving, and path-finding for NASA's Mars rovers. However, two key problems remain challenging in OLfD: 1) heterogeneity: demons...
Sravan Jayanthi;Letian Chen;Nadya Balabanska;Van Duong;Erik Scarlatescu;Ezra Ameperosa;Zulfiqar Haider Zaidi;Daniel Martin;Taylor Keith Del Matto;Masahiro Ono;Matthew Gombolay
Georgia Institute of Technology;Toyota Research Institute;Jet Propulsion Laboratory;;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Jet Propulsion Laboratory;Georgia Institute of Technology
Poster
main
Learning from Heterogeneous Demonstration;Network Distillation;Offline Imitation Learning
https://openreview.net/forum?id=h-geaPzuJu
5
DROID: Learning from Offline Heterogeneous Demonstrations via Reward-Policy Distillation Offline Learning from Demonstrations (OLfD) is valuable in domains where trial-and-error learning is infeasible or specifying a cost function is difficult, such as robotic surgery, autonomous driving, and path-finding for NASA's Ma...
[ -0.060748882591724396, -0.03408045694231987, -0.04227466508746147, -0.01690986566245556, -0.03862451761960983, -0.023427987471222878, -0.02672429382801056, 0.007467903662472963, 0.01727301813662052, 0.03126835450530052, -0.04491915926337242, -0.03558893874287605, -0.002544394927099347, 0.0...
corl_2023_h8halpbqB-
h8halpbqB-
corl
2,023
Im2Contact: Vision-Based Contact Localization Without Touch or Force Sensing
Contacts play a critical role in most manipulation tasks. Robots today mainly use proximal touch/force sensors to sense contacts, but the information they provide must be calibrated and is inherently local, with practical applications relying either on extensive surface coverage or restrictive assumptions to resolve am...
Leon Kim;Yunshuang Li;Michael Posa;Dinesh Jayaraman
University of Pennsylvania;University of Pennsylvania;University of Pennsylvania;University of Pennsylvania
Poster
main
contact perception;manipulation;vision-based
https://openreview.net/forum?id=h8halpbqB-
4
Im2Contact: Vision-Based Contact Localization Without Touch or Force Sensing Contacts play a critical role in most manipulation tasks. Robots today mainly use proximal touch/force sensors to sense contacts, but the information they provide must be calibrated and is inherently local, with practical applications relying ...
[ -0.026002639904618263, -0.008147739805281162, 0.0010651808697730303, 0.026759281754493713, -0.04004665091633797, -0.007035845424979925, 0.0247846320271492, -0.03369823843240738, 0.02068769372999668, 0.007875533774495125, -0.028992297127842903, 0.00516269588842988, 0.04240884631872177, 0.00...
corl_2023_hRZ1YjDZmTo
hRZ1YjDZmTo
corl
2,023
MimicPlay: Long-Horizon Imitation Learning by Watching Human Play
Imitation learning from human demonstrations is a promising paradigm for teaching robots manipulation skills in the real world. However, learning complex long-horizon tasks often requires an unattainable amount of demonstrations. To reduce the high data requirement, we resort to human play data - video sequences of peo...
Chen Wang;Linxi Fan;Jiankai Sun;Ruohan Zhang;Li Fei-Fei;Danfei Xu;Yuke Zhu;Anima Anandkumar
Computer Science Department, Stanford University;NVIDIA;Stanford University;Stanford University;Stanford University;NVIDIA;Computer Science Department, University of Texas, Austin;California Institute of Technology
Oral
main
Imitation Learning;Learning from Human;Long-Horizon Manipulation
https://github.com/j96w/MimicPlay
https://openreview.net/forum?id=hRZ1YjDZmTo
187
MimicPlay: Long-Horizon Imitation Learning by Watching Human Play Imitation learning from human demonstrations is a promising paradigm for teaching robots manipulation skills in the real world. However, learning complex long-horizon tasks often requires an unattainable amount of demonstrations. To reduce the high data ...
[ -0.04118998721241951, -0.039260365068912506, -0.03139344975352287, 0.0017174091190099716, -0.013275421224534512, -0.018776696175336838, -0.014852515421807766, -0.00950894970446825, -0.010881949216127396, 0.004566614981740713, -0.029241180047392845, -0.025344830006361008, -0.01903645321726799...
corl_2023_i84V7i6KEMd
i84V7i6KEMd
corl
2,023
Sample-Efficient Preference-based Reinforcement Learning with Dynamics Aware Rewards
Preference-based reinforcement learning (PbRL) aligns a robot behavior with human preferences via a reward function learned from binary feedback over agent behaviors. We show that encoding environment dynamics in the reward function improves the sample efficiency of PbRL by an order of magnitude. In our experiments we ...
Katherine Metcalf;Miguel Sarabia;Natalie Mackraz;Barry-John Theobald
;;Apple;Apple
Poster
main
human-in-the-loop learning;preference-based RL;RLHF
https://openreview.net/forum?id=i84V7i6KEMd
7
Sample-Efficient Preference-based Reinforcement Learning with Dynamics Aware Rewards Preference-based reinforcement learning (PbRL) aligns a robot behavior with human preferences via a reward function learned from binary feedback over agent behaviors. We show that encoding environment dynamics in the reward function im...
[ -0.0683978945016861, -0.020455794408917427, -0.005352350417524576, 0.006736014503985643, -0.04405288025736809, -0.026775775477290154, 0.026345716789364815, -0.015566224232316017, 0.022568687796592712, 0.014332144521176815, -0.04289359599351883, 0.0004654090735130012, -0.010293341241776943, ...
corl_2023_ihqTtzS83VS
ihqTtzS83VS
corl
2,023
Learning Reusable Manipulation Strategies
Humans demonstrate an impressive ability to acquire and generalize manipulation "tricks." Even from a single demonstration, such as using soup ladles to reach for distant objects, we can apply this skill to new scenarios involving different object positions, sizes, and categories (e.g., forks and hammers). Additionally...
Jiayuan Mao;Tomás Lozano-Pérez;Joshua B. Tenenbaum;Leslie Pack Kaelbling
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Contact Modeling and Manipulation;Task and Motion Planning
https://openreview.net/forum?id=ihqTtzS83VS
11
Learning Reusable Manipulation Strategies Humans demonstrate an impressive ability to acquire and generalize manipulation "tricks." Even from a single demonstration, such as using soup ladles to reach for distant objects, we can apply this skill to new scenarios involving different object positions, sizes, and categori...
[ -0.00021476828260347247, -0.004084956832230091, -0.0269695483148098, 0.012424591928720474, -0.0005574094248004258, -0.01062972191721201, 0.0012322294060140848, 0.00540320947766304, -0.01770690456032753, 0.022598622366786003, -0.03152647241950035, 0.0034874416887760162, -0.0024086597841233015...
corl_2023_j2AQ-WJ_ze
j2AQ-WJ_ze
corl
2,023
Language-guided Robot Grasping: CLIP-based Referring Grasp Synthesis in Clutter
Robots operating in human-centric environments require the integration of visual grounding and grasping capabilities to effectively manipulate objects based on user instructions. This work focuses on the task of referring grasp synthesis, which predicts a grasp pose for an object referred through natural language in cl...
Georgios Tziafas;Yucheng XU;Arushi Goel;Mohammadreza Kasaei;Zhibin Li;Hamidreza Kasaei
University of Groningen;University of Edinburgh, University of Edinburgh;University of Edinburgh;University of Edinburgh, University of Edinburgh;University College London, University of London;University of Groningen
Poster
main
Language-Guided Robot Grasping;Referring Grasp Synthesis;Visual Grounding
https://github.com/gtziafas/OCID-VLG
https://openreview.net/forum?id=j2AQ-WJ_ze
27
Language-guided Robot Grasping: CLIP-based Referring Grasp Synthesis in Clutter Robots operating in human-centric environments require the integration of visual grounding and grasping capabilities to effectively manipulate objects based on user instructions. This work focuses on the task of referring grasp synthesis, w...
[ -0.05382906645536423, -0.010271729901432991, -0.03114953264594078, 0.009565897285938263, -0.02483418397605419, -0.0360160656273365, -0.006398934870958328, -0.02944067306816578, -0.018398098647594452, 0.001026824815198779, -0.02423979714512825, 0.0009391760686412454, 0.03616466373205185, -0...
corl_2023_k-Fg8JDQmc
k-Fg8JDQmc
corl
2,023
Language Embedded Radiance Fields for Zero-Shot Task-Oriented Grasping
Grasping objects by a specific subpart is often crucial for safety and for executing downstream tasks. We propose LERF-TOGO, Language Embedded Radiance Fields for Task-Oriented Grasping of Objects, which uses vision-language models zero-shot to output a grasp distribution over an object given a natural language query. ...
Adam Rashid;Satvik Sharma;Chung Min Kim;Justin Kerr;Lawrence Yunliang Chen;Angjoo Kanazawa;Ken Goldberg
;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley
Oral
main
NeRF;Natural Language;Grasping;Semantics
https://openreview.net/forum?id=k-Fg8JDQmc
88
Language Embedded Radiance Fields for Zero-Shot Task-Oriented Grasping Grasping objects by a specific subpart is often crucial for safety and for executing downstream tasks. We propose LERF-TOGO, Language Embedded Radiance Fields for Task-Oriented Grasping of Objects, which uses vision-language models zero-shot to outp...
[ -0.04120490327477455, -0.012317084707319736, 0.012612991034984589, 0.002151097636669874, 0.0033844243735074997, -0.008997390046715736, 0.018577344715595245, 0.0008501516422256827, -0.025336947292089462, -0.020251063629984856, -0.026945935562253, -0.005807154346257448, -0.027241840958595276, ...
corl_2023_kOm3jWX8YN
kOm3jWX8YN
corl
2,023
Learning to Discern: Imitating Heterogeneous Human Demonstrations with Preference and Representation Learning
Practical Imitation Learning (IL) systems rely on large human demonstration datasets for successful policy learning. However, challenges lie in maintaining the quality of collected data and addressing the suboptimal nature of some demonstrations, which can compromise the overall dataset quality and hence the learning o...
Sachit Kuhar;Shuo Cheng;Shivang Chopra;Matthew Bronars;Danfei Xu
Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;NVIDIA
Poster
main
Imitation Learning;Preference Learning;Manipulation
https://openreview.net/forum?id=kOm3jWX8YN
9
Learning to Discern: Imitating Heterogeneous Human Demonstrations with Preference and Representation Learning Practical Imitation Learning (IL) systems rely on large human demonstration datasets for successful policy learning. However, challenges lie in maintaining the quality of collected data and addressing the subop...
[ -0.07097481191158295, -0.03932541608810425, 0.00361113203689456, 0.015739619731903076, -0.030779700726270676, 0.002798154717311263, -0.01861339993774891, -0.0057806470431387424, 0.0010114951292052865, 0.011788171716034412, -0.05698404088616371, -0.013054903596639633, -0.023916659876704216, ...
corl_2023_kSXh83gWWy
kSXh83gWWy
corl
2,023
Context-Aware Deep Reinforcement Learning for Autonomous Robotic Navigation in Unknown Area
Mapless navigation refers to a challenging task where a mobile robot must rapidly navigate to a predefined destination using its partial knowledge of the environment, which is updated online along the way, instead of a prior map of the environment. Inspired by the recent developments in deep reinforcement learning (DRL...
Jingsong Liang;Zhichen Wang;Yuhong Cao;Jimmy Chiun;Mengqi Zhang;Guillaume Adrien Sartoretti
National University of Singapore, NUS;National University of Singapore;National University of Singapore;National University of Singapore;National University of Singapore;National University of Singapore
Poster
main
deep reinforcement learning;mapless navigation;context-aware decision-making
https://github.com/marmotlab/Context_Aware_Navigation
https://openreview.net/forum?id=kSXh83gWWy
10
Context-Aware Deep Reinforcement Learning for Autonomous Robotic Navigation in Unknown Area Mapless navigation refers to a challenging task where a mobile robot must rapidly navigate to a predefined destination using its partial knowledge of the environment, which is updated online along the way, instead of a prior map...
[ -0.0629289299249649, -0.017391672357916832, -0.005302791483700275, 0.008705106563866138, -0.031668417155742645, -0.03296630457043648, 0.027181440964341164, -0.008913695812225342, 0.030277825891971588, 0.021841567009687424, -0.03550645336508751, -0.021804483607411385, 0.015889832749962807, ...
corl_2023_keAPCON4jHC
keAPCON4jHC
corl
2,023
Robust Reinforcement Learning in Continuous Control Tasks with Uncertainty Set Regularization
Reinforcement learning (RL) is recognized as lacking generalization and robustness under environmental perturbations, which excessively restricts its application for real-world robotics. Prior work claimed that adding regularization to the value function is equivalent to learning a robust policy under uncertain transit...
Yuan Zhang;Jianhong Wang;Joschka Boedecker
University of Freiburg;Imperial College London;Universität Freiburg
Poster
main
Reinforcement Learning;Robustness;Continuous Control;Robotics
github.com/mikezhang95/rrl_usr
https://openreview.net/forum?id=keAPCON4jHC
5
Robust Reinforcement Learning in Continuous Control Tasks with Uncertainty Set Regularization Reinforcement learning (RL) is recognized as lacking generalization and robustness under environmental perturbations, which excessively restricts its application for real-world robotics. Prior work claimed that adding regulari...
[ -0.07754991948604584, 0.027371663600206375, -0.00858534686267376, -0.0073313601315021515, -0.020138932392001152, -0.0032523812260478735, -0.009815850295126438, 0.05511905252933502, 0.020514657720923424, -0.0014077996602281928, -0.0322561077773571, -0.016212591901421547, 0.011412687599658966,...
corl_2023_low-53sFqn
low-53sFqn
corl
2,023
Fleet Active Learning: A Submodular Maximization Approach
In multi-robot systems, robots often gather data to improve the performance of their deep neural networks (DNNs) for perception and planning. Ideally, these robots should select the most informative samples from their local data distributions by employing active learning approaches. However, when the data collection is...
Oguzhan Akcin;Orhan Unuvar;Onat Ure;Sandeep P. Chinchali
The University of Texas at Austin;University of Texas at Austin;University of Texas at Austin;University of Texas at Austin
Poster
main
Active Learning;Cloud Robotics;Robotic Perception
https://github.com/UTAustin-SwarmLab/Fleet-Active-Learning.git
https://openreview.net/forum?id=low-53sFqn
5
Fleet Active Learning: A Submodular Maximization Approach In multi-robot systems, robots often gather data to improve the performance of their deep neural networks (DNNs) for perception and planning. Ideally, these robots should select the most informative samples from their local data distributions by employing active...
[ -0.029853245243430138, -0.08140799403190613, -0.029871482402086258, 0.029196729883551598, -0.035105373710393906, -0.03463122621178627, 0.036837849766016006, -0.006510451436042786, -0.001170559087768197, 0.05357901006937027, -0.008890322409570217, -0.015537547878921032, -0.01850098930299282, ...
corl_2023_mTZcxs2O7k
mTZcxs2O7k
corl
2,023
Batch Differentiable Pose Refinement for In-The-Wild Camera/LiDAR Extrinsic Calibration
Accurate camera to LiDAR (Light Detection and Ranging) extrinsic calibration is important for robotic tasks carrying out tight sensor fusion --- such as target tracking and odometry. Calibration is typically performed before deployment in controlled conditions using calibration targets, however, this limits scalability...
Lanke Frank Tarimo Fu;Maurice Fallon
University of Oxford;University of Oxford
Poster
main
Sensor Fusion;Extrinsic Calibration;Differentiable Optimization
https://openreview.net/forum?id=mTZcxs2O7k
6
Batch Differentiable Pose Refinement for In-The-Wild Camera/LiDAR Extrinsic Calibration Accurate camera to LiDAR (Light Detection and Ranging) extrinsic calibration is important for robotic tasks carrying out tight sensor fusion --- such as target tracking and odometry. Calibration is typically performed before deploym...
[ -0.045777130872011185, 0.0002851674216799438, -0.057403646409511566, 0.01712086610496044, -0.02376328781247139, -0.007981840521097183, 0.044829513877630234, -0.011763191781938076, 0.013895327225327492, 0.03797752410173416, -0.030797511339187622, 0.01004108227789402, -0.00885656289756298, 0...
corl_2023_n9lew97SAn
n9lew97SAn
corl
2,023
Action-Quantized Offline Reinforcement Learning for Robotic Skill Learning
The offline reinforcement learning (RL) paradigm provides a general recipe to convert static behavior datasets into policies that can perform better than the policy that collected the data. While policy constraints, conservatism, and other methods for mitigating distributional shifts have made offline reinforcement lea...
Jianlan Luo;Perry Dong;Jeffrey Wu;Aviral Kumar;Xinyang Geng;Sergey Levine
Google;;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Google
Poster
main
Offline Reinforcement Learning;Discretization;Robot Skill Learning
https://openreview.net/forum?id=n9lew97SAn
25
Action-Quantized Offline Reinforcement Learning for Robotic Skill Learning The offline reinforcement learning (RL) paradigm provides a general recipe to convert static behavior datasets into policies that can perform better than the policy that collected the data. While policy constraints, conservatism, and other metho...
[ -0.0690653994679451, 0.004637713544070721, 0.0034858479630202055, -0.00203147204592824, -0.024200813844799995, 0.009475549682974815, 0.04196048900485039, 0.05071001127362251, 0.02153872326016426, 0.040359511971473694, -0.028854817152023315, -0.020123906433582306, -0.045646458864212036, 0.0...
corl_2023_nKWQnYkkwX
nKWQnYkkwX
corl
2,023
Language-Guided Traffic Simulation via Scene-Level Diffusion
Realistic and controllable traffic simulation is a core capability that is necessary to accelerate autonomous vehicle (AV) development. However, current approaches for controlling learning-based traffic models require significant domain expertise and are difficult for practitioners to use. To remedy this, we present CT...
Ziyuan Zhong;Davis Rempe;Yuxiao Chen;Boris Ivanovic;Yulong Cao;Danfei Xu;Marco Pavone;Baishakhi Ray
Columbia University;Stanford University;NVIDIA;NVIDIA;NVIDIA;Stanford University;Columbia University;California Institute of Technology
Oral
main
Traffic Simulation;Multi-Agent Diffusion;Large Language Model
https://openreview.net/forum?id=nKWQnYkkwX
94
Language-Guided Traffic Simulation via Scene-Level Diffusion Realistic and controllable traffic simulation is a core capability that is necessary to accelerate autonomous vehicle (AV) development. However, current approaches for controlling learning-based traffic models require significant domain expertise and are diff...
[ -0.08106561005115509, -0.04112398624420166, 0.010465740226209164, -0.0023716494906693697, -0.01565704308450222, -0.024848051369190216, -0.015370690263807774, 0.04954830929636955, -0.004969610366970301, 0.015269081108272076, -0.02826581336557865, -0.024755679070949554, 0.005606976803392172, ...
corl_2023_nNsZxc2cmO
nNsZxc2cmO
corl
2,023
FindThis: Language-Driven Object Disambiguation in Indoor Environments
Natural language is naturally ambiguous. In this work, we consider interactions between a user and a mobile service robot tasked with locating a desired object, specified by a language utterance. We present a task FindThis, which addresses the problem of how to disambiguate and locate the particular object instance des...
Arjun Majumdar;Fei Xia;brian ichter;Dhruv Batra;Leonidas Guibas
Georgia Institute of Technology;Google;Google;Georgia Institute of Technology;Stanford University
Poster
main
object disambiguation;instruction following;language interaction;visual navigation
https://openreview.net/forum?id=nNsZxc2cmO
11
FindThis: Language-Driven Object Disambiguation in Indoor Environments Natural language is naturally ambiguous. In this work, we consider interactions between a user and a mobile service robot tasked with locating a desired object, specified by a language utterance. We present a task FindThis, which addresses the probl...
[ -0.04861170053482056, -0.01916060969233513, -0.006382166873663664, 0.006819559261202812, -0.02031758427619934, -0.01395893283188343, 0.000059854861319763586, -0.03442701697349548, 0.024287037551403046, 0.006513854954391718, -0.03946878761053085, -0.040070790797472, -0.011315765790641308, -...
corl_2023_nyY6UgXYyfF
nyY6UgXYyfF
corl
2,023
Adv3D: Generating Safety-Critical 3D Objects through Closed-Loop Simulation
Self-driving vehicles (SDVs) must be rigorously tested on a wide range of scenarios to ensure safe deployment. The industry typically relies on closed-loop simulation to evaluate how the SDV interacts on a corpus of synthetic and real scenarios and to verify good performance. However, they primarily only test the motio...
Jay Sarva;Jingkang Wang;James Tu;Yuwen Xiong;Sivabalan Manivasagam;Raquel Urtasun
Brown University;University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;;Department of Computer Science, University of Toronto
Poster
main
closed-loop simulation;adversarial robustness;self-driving
https://openreview.net/forum?id=nyY6UgXYyfF
7
Adv3D: Generating Safety-Critical 3D Objects through Closed-Loop Simulation Self-driving vehicles (SDVs) must be rigorously tested on a wide range of scenarios to ensure safe deployment. The industry typically relies on closed-loop simulation to evaluate how the SDV interacts on a corpus of synthetic and real scenarios...
[ -0.05939937382936478, 0.002845299895852804, -0.015999924391508102, 0.012706093490123749, -0.02600102685391903, 0.025467390194535255, 0.015263873152434826, -0.004524418152868748, 0.0010707252658903599, 0.020002206787467003, -0.0004194919893052429, -0.0019562875386327505, -0.005676798988133669...
corl_2023_o-K3HVUeEw
o-K3HVUeEw
corl
2,023
Composable Part-Based Manipulation
In this paper, we propose composable part-based manipulation (CPM), a novel approach that leverages object-part decomposition and part-part correspondences to improve learning and generalization of robotic manipulation skills. By considering the functional correspondences between object parts, we conceptualize function...
Weiyu Liu;Jiayuan Mao;Joy Hsu;Tucker Hermans;Animesh Garg;Jiajun Wu
Stanford University;Massachusetts Institute of Technology;Stanford University;University of Utah;University of Toronto;Stanford University
Poster
main
Manipulation;Part Decomposition;Diffusion Model
https://cpmcorl2023.github.io/
https://openreview.net/forum?id=o-K3HVUeEw
14
Composable Part-Based Manipulation In this paper, we propose composable part-based manipulation (CPM), a novel approach that leverages object-part decomposition and part-part correspondences to improve learning and generalization of robotic manipulation skills. By considering the functional correspondences between obje...
[ -0.011886194348335266, -0.016953807324171066, -0.017611578106880188, 0.02397620491683483, 0.012367941439151764, -0.02406884916126728, 0.021585995331406593, -0.004312566947191954, 0.028663979843258858, 0.04395020008087158, -0.03650164231657982, 0.004761889111250639, 0.027033448219299316, 0....
corl_2023_o2wNSCTkq0
o2wNSCTkq0
corl
2,023
Learning Sequential Acquisition Policies for Robot-Assisted Feeding
A robot providing mealtime assistance must perform specialized maneuvers with various utensils in order to pick up and feed a range of food items. Beyond these dexterous low-level skills, an assistive robot must also plan these strategies in sequence over a long horizon to clear a plate and complete a meal. Previous me...
Priya Sundaresan;Jiajun Wu;Dorsa Sadigh
Stanford University;Stanford University;Stanford University
Poster
main
Deformable Manipulation;Dexterous Manipulation
https://sites.google.com/view/vaporsbot
https://openreview.net/forum?id=o2wNSCTkq0
10
Learning Sequential Acquisition Policies for Robot-Assisted Feeding A robot providing mealtime assistance must perform specialized maneuvers with various utensils in order to pick up and feed a range of food items. Beyond these dexterous low-level skills, an assistive robot must also plan these strategies in sequence o...
[ -0.03195144608616829, -0.0317835733294487, -0.009251556359231472, 0.050398603081703186, -0.055136293172836304, 0.011284660547971725, 0.004352988209575415, 0.01674979366362095, -0.005950093735009432, 0.038069408386945724, 0.0025250595062971115, -0.04069938883185387, 0.009340154938399792, 0....
corl_2023_o82EXEK5hu6
o82EXEK5hu6
corl
2,023
Parting with Misconceptions about Learning-based Vehicle Motion Planning
The release of nuPlan marks a new era in vehicle motion planning research, offering the first large-scale real-world dataset and evaluation schemes requiring both precise short-term planning and long-horizon ego-forecasting. Existing systems struggle to simultaneously meet both requirements. Indeed, we find that these ...
Daniel Dauner;Marcel Hallgarten;Andreas Geiger;Kashyap Chitta
Eberhard-Karls-Universität Tübingen;Eberhard-Karls-Universität Tübingen;University of Tuebingen;University of Tübingen
Poster
main
Motion Planning;Autonomous Driving;Data-driven Simulation
https://github.com/autonomousvision/tuplan_garage
https://openreview.net/forum?id=o82EXEK5hu6
141
Parting with Misconceptions about Learning-based Vehicle Motion Planning The release of nuPlan marks a new era in vehicle motion planning research, offering the first large-scale real-world dataset and evaluation schemes requiring both precise short-term planning and long-horizon ego-forecasting. Existing systems strug...
[ -0.08334004878997803, -0.03483673557639122, -0.04520769417285919, 0.031243208795785904, -0.03100115805864334, 0.04018047824501991, -0.0040683201514184475, 0.02040676772594452, -0.01036164816468954, 0.024093391373753548, 0.004077629651874304, -0.018852055072784424, 0.010882988572120667, 0.0...
corl_2023_oqOfLP6bJy
oqOfLP6bJy
corl
2,023
Contrastive Value Learning: Implicit Models for Simple Offline RL
Model-based reinforcement learning (RL) methods are appealing in the offline setting because they allow an agent to reason about the consequences of actions without interacting with the environment. While conventional model-based methods learn a 1-step model, predicting the immediate next state, these methods must be p...
Bogdan Mazoure;Benjamin Eysenbach;Ofir Nachum;Jonathan Tompson
Apple;Carnegie Mellon University;OpenAI;Google DeepMind
Poster
main
reinforcement learning;robotics;metaworld;unsupervised learning;contrastive learning;noise-contrastive estimation;generative model
https://openreview.net/forum?id=oqOfLP6bJy
11
Contrastive Value Learning: Implicit Models for Simple Offline RL Model-based reinforcement learning (RL) methods are appealing in the offline setting because they allow an agent to reason about the consequences of actions without interacting with the environment. While conventional model-based methods learn a 1-step m...
[ -0.08270052075386047, -0.025008460506796837, -0.010071322321891785, 0.023704789578914642, -0.009520474821329117, -0.0008847982389852405, 0.030241509899497032, 0.010025418363511562, 0.018297307193279266, -0.02341100387275219, -0.02497173845767975, -0.03457484021782875, 0.008629938587546349, ...
corl_2023_oyWkrG-LD5
oyWkrG-LD5
corl
2,023
Geometry Matching for Multi-Embodiment Grasping
While significant progress has been made on the problem of generating grasps, many existing learning-based approaches still concentrate on a single embodiment, provide limited generalization to higher DoF end-effectors and cannot capture a diverse set of grasp modes. In this paper, we tackle the problem of grasping mul...
Maria Attarian;Muhammad Adil Asif;Jingzhou Liu;Ruthrash Hari;Animesh Garg;Igor Gilitschenski;Jonathan Tompson
Google;University of Toronto;University of Toronto;;University of Toronto;University of Toronto;Google DeepMind
Poster
main
Multi-Embodiment;Dexterous Grasping;Graph Neural Networks
https://openreview.net/forum?id=oyWkrG-LD5
9
Geometry Matching for Multi-Embodiment Grasping While significant progress has been made on the problem of generating grasps, many existing learning-based approaches still concentrate on a single embodiment, provide limited generalization to higher DoF end-effectors and cannot capture a diverse set of grasp modes. In t...
[ -0.039434585720300674, -0.03720695525407791, 0.0031251264736056328, -0.03076339326798916, -0.03184959292411804, -0.04256431385874748, 0.022368356585502625, -0.0023749119136482477, -0.005490833427757025, -0.012546530924737453, -0.042527493089437485, 0.0004619802057277411, 0.003859231946989894...
corl_2023_pLCQkMojXI
pLCQkMojXI
corl
2,023
Rearrangement Planning for General Part Assembly
Most successes in autonomous robotic assembly have been restricted to single target or category. We propose to investigate general part assembly, the task of creating novel target assemblies with unseen part shapes. As a fundamental step to a general part assembly system, we tackle the task of determining the precise p...
Yulong Li;Andy Zeng;Shuran Song
Columbia University;Columbia University;Google
Oral
main
robotic assembly;pose estimation;3D perception
https://github.com/real-stanford/gpat
https://openreview.net/forum?id=pLCQkMojXI
11
Rearrangement Planning for General Part Assembly Most successes in autonomous robotic assembly have been restricted to single target or category. We propose to investigate general part assembly, the task of creating novel target assemblies with unseen part shapes. As a fundamental step to a general part assembly system...
[ -0.013152917847037315, -0.0018244963139295578, -0.026636727154254913, 0.03652669116854668, 0.0029734233394265175, -0.015689749270677567, 0.009090311825275421, -0.02720659412443638, -0.0026057667564600706, 0.06180308759212494, 0.0008249296806752682, -0.025919796898961067, -0.00563893420621752...
corl_2023_psLlVbTFBua
psLlVbTFBua
corl
2,023
FlowBot++: Learning Generalized Articulated Objects Manipulation via Articulation Projection
Understanding and manipulating articulated objects, such as doors and drawers, is crucial for robots operating in human environments. We wish to develop a system that can learn to articulate novel objects with no prior interaction, after training on other articulated objects. Previous approaches for articulated object...
Harry Zhang;Ben Eisner;David Held
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Articulated objects manipulation;representation learning
https://sites.google.com/view/flowbotpp/home
https://openreview.net/forum?id=psLlVbTFBua
35
FlowBot++: Learning Generalized Articulated Objects Manipulation via Articulation Projection Understanding and manipulating articulated objects, such as doors and drawers, is crucial for robots operating in human environments. We wish to develop a system that can learn to articulate novel objects with no prior interact...
[ -0.030265264213085175, -0.021015534177422523, -0.018282247707247734, 0.044275619089603424, -0.004873757250607014, -0.02154046855866909, -0.011476182378828526, -0.01703326217830181, -0.0005469966563396156, 0.02260844223201275, -0.012399345636367798, -0.013449217192828655, 0.01575712487101555,...
corl_2023_psyvs5wdAV
psyvs5wdAV
corl
2,023
Equivariant Motion Manifold Primitives
Existing movement primitive models for the most part focus on representing and generating a single trajectory for a given task, limiting their adaptability to situations in which unforeseen obstacles or new constraints may arise. In this work we propose Motion Manifold Primitives (MMP), a movement primitive paradigm th...
Byeongho Lee;Yonghyeon Lee;Seungyeon Kim;MinJun Son;Frank C. Park
;Seoul National University;Seoul National University;Seoul National University;Seoul National University
Poster
main
Movement primitives;Manifold;LfD;Equivariance
https://github.com/dlsfldl/EMMP-public
https://openreview.net/forum?id=psyvs5wdAV
9
Equivariant Motion Manifold Primitives Existing movement primitive models for the most part focus on representing and generating a single trajectory for a given task, limiting their adaptability to situations in which unforeseen obstacles or new constraints may arise. In this work we propose Motion Manifold Primitives ...
[ -0.06676073372364044, -0.020615626126527786, -0.021707594394683838, 0.017170274630188942, -0.013056562282145023, -0.01663370244204998, 0.007460220716893673, 0.008698099292814732, 0.02765694446861744, 0.001230818685144186, -0.049891695380210876, -0.02874891273677349, -0.016793733462691307, ...
corl_2023_pw-OTIYrGa
pw-OTIYrGa
corl
2,023
On the Utility of Koopman Operator Theory in Learning Dexterous Manipulation Skills
Despite impressive dexterous manipulation capabilities enabled by learning-based approaches, we are yet to witness widespread adoption beyond well-resourced laboratories. This is likely due to practical limitations, such as significant computational burden, inscrutable learned behaviors, sensitivity to initialization, ...
Yunhai Han;Mandy Xie;Ye Zhao;Harish Ravichandar
Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology
Oral
main
Koopman Operator;Dexterous Manipulation
https://github.com/GT-STAR-Lab/KODex
https://openreview.net/forum?id=pw-OTIYrGa
16
On the Utility of Koopman Operator Theory in Learning Dexterous Manipulation Skills Despite impressive dexterous manipulation capabilities enabled by learning-based approaches, we are yet to witness widespread adoption beyond well-resourced laboratories. This is likely due to practical limitations, such as significant ...
[ -0.020192066207528114, -0.001411694334819913, -0.054128002375364304, 0.01038159616291523, -0.020099949091672897, -0.018128642812371254, 0.036349404603242874, 0.01973148062825203, 0.015282224863767624, 0.02450314536690712, -0.053575299680233, -0.022421298548579216, 0.029145846143364906, 0.0...
corl_2023_q0VAoefCI2
q0VAoefCI2
corl
2,023
Task-Oriented Koopman-Based Control with Contrastive Encoder
We present task-oriented Koopman-based control that utilizes end-to-end reinforcement learning and contrastive encoder to simultaneously learn the Koopman latent embedding, operator, and associated linear controller within an iterative loop. By prioritizing the task cost as the main objective for controller learning, w...
Xubo Lyu;Hanyang Hu;Seth Siriya;Ye Pu;Mo Chen
Simon Fraser University;Simon Fraser University;University of Melbourne;University of Melbourne;Simon Fraser University
Oral
main
Learning and control;Koopman-based control;Represention learning
https://sites.google.com/view/kpmlilatsupp/
https://openreview.net/forum?id=q0VAoefCI2
6
Task-Oriented Koopman-Based Control with Contrastive Encoder We present task-oriented Koopman-based control that utilizes end-to-end reinforcement learning and contrastive encoder to simultaneously learn the Koopman latent embedding, operator, and associated linear controller within an iterative loop. By prioritizing t...
[ -0.03497523441910744, -0.0029331082478165627, -0.027313990518450737, -0.0005337655311450362, -0.007772273849695921, -0.007527077104896307, 0.031144611537456512, 0.0031042832415550947, -0.00018028321210294962, 0.018033526837825775, -0.031477708369493484, -0.00872067641466856, 0.01290752645581...
corl_2023_qVc7NWYTRZ6
qVc7NWYTRZ6
corl
2,023
An Unbiased Look at Datasets for Visuo-Motor Pre-Training
Visual representation learning hold great promise for robotics, but is severely hampered by the scarcity and homogeneity of robotics datasets. Recent works address this problem by pre-training visual representations on large-scale but out-of-domain data (e.g., videos of egocentric interactions) and then transferring th...
Sudeep Dasari;Mohan Kumar Srirama;Unnat Jain;Abhinav Gupta
;Carnegie Mellon University;;Carnegie Mellon University
Poster
main
Visual Representation Learning;Datasets;Manipulation
https://github.com/SudeepDasari/data4robotics
https://openreview.net/forum?id=qVc7NWYTRZ6
37
An Unbiased Look at Datasets for Visuo-Motor Pre-Training Visual representation learning hold great promise for robotics, but is severely hampered by the scarcity and homogeneity of robotics datasets. Recent works address this problem by pre-training visual representations on large-scale but out-of-domain data (e.g., v...
[ -0.04920327663421631, -0.02896704152226448, -0.027931183576583862, 0.016074303537607193, -0.052458830177783966, -0.0030728925485163927, 0.018793432042002678, 0.03538566455245018, 0.008014029823243618, 0.027043305337429047, -0.02267790026962757, -0.04468989372253418, 0.020051259547472, 0.00...
corl_2023_rOCWUmMBSnH
rOCWUmMBSnH
corl
2,023
A Policy Optimization Method Towards Optimal-time Stability
In current model-free reinforcement learning (RL) algorithms, stability criteria based on sampling methods are commonly utilized to guide policy optimization. However, these criteria only guarantee the infinite-time convergence of the system's state to an equilibrium point, which leads to sub-optimality of the policy. ...
Shengjie Wang;Lan Fengb;Xiang Zheng;Yuxue Cao;Oluwatosin OluwaPelumi Oseni;Haotian Xu;Tao Zhang;Yang Gao
Tsinghua University;Tsinghua University;;;;;;Tsinghua University
Poster
main
Reinforcement Learning;Robotic Control;Stability
https://openreview.net/forum?id=rOCWUmMBSnH
3
A Policy Optimization Method Towards Optimal-time Stability In current model-free reinforcement learning (RL) algorithms, stability criteria based on sampling methods are commonly utilized to guide policy optimization. However, these criteria only guarantee the infinite-time convergence of the system's state to an equi...
[ -0.02096559666097164, -0.032122619450092316, -0.036038659512996674, 0.014084817841649055, -0.015017648227512836, 0.00352813140489161, 0.03018307127058506, 0.023108337074518204, 0.01534090656787157, 0.0474727638065815, -0.032067205756902695, -0.016550814732909203, -0.008926541544497013, 0.0...
corl_2023_rPye6EZxmI
rPye6EZxmI
corl
2,023
Reinforcement Learning Enables Real-Time Planning and Control of Agile Maneuvers for Soft Robot Arms
Control policies for soft robot arms typically assume quasi-static motion or require a hand-designed motion plan. To achieve real-time planning and control for tasks requiring highly dynamic maneuvers, we apply deep reinforcement learning to train a policy entirely in simulation, and we identify strategies and insights...
Rianna Jitosho;Tyler Ga Wei Lum;Allison Okamura;Karen Liu
Stanford University;Stanford University;;Computer Science Department, Stanford University
Poster
main
Soft Robotics;Reinforcement Learning;Sim-to-Real Transfer;Dynamics and Control
https://github.com/tylerlum/Vine_Robot_IsaacGymEnvs
https://openreview.net/forum?id=rPye6EZxmI
13
Reinforcement Learning Enables Real-Time Planning and Control of Agile Maneuvers for Soft Robot Arms Control policies for soft robot arms typically assume quasi-static motion or require a hand-designed motion plan. To achieve real-time planning and control for tasks requiring highly dynamic maneuvers, we apply deep rei...
[ -0.06809594482183456, -0.01709744520485401, -0.0039805276319384575, -0.0030990266241133213, -0.02844676934182644, -0.029658833518624306, 0.030209770426154137, 0.032597169280052185, 0.020403074100613594, 0.014563130214810371, -0.022937389090657234, -0.004728885367512703, -0.009324626997113228...
corl_2023_rYZBdBytxBx
rYZBdBytxBx
corl
2,023
HOI4ABOT: Human-Object Interaction Anticipation for Human Intention Reading Collaborative roBOTs
Robots are becoming increasingly integrated into our lives, assisting us in various tasks. To ensure effective collaboration between humans and robots, it is essential that they understand our intentions and anticipate our actions. In this paper, we propose a Human-Object Interaction (HOI) anticipation framework for co...
Esteve Valls Mascaro;Daniel Sliwowski;Dongheui Lee
Technische Universität Wien;Technische Universität Wien;Technische Universität Wien
Poster
main
Human-Object Interaction;Human-Robot Collaboration;Human Intention
https://openreview.net/forum?id=rYZBdBytxBx
10
HOI4ABOT: Human-Object Interaction Anticipation for Human Intention Reading Collaborative roBOTs Robots are becoming increasingly integrated into our lives, assisting us in various tasks. To ensure effective collaboration between humans and robots, it is essential that they understand our intentions and anticipate our ...
[ -0.013913167640566826, -0.04400785639882088, -0.028450820595026016, 0.03443853557109833, -0.018073342740535736, -0.018358035013079643, -0.01897333562374115, -0.003184416564181447, 0.004454050213098526, 0.022830452769994736, -0.044228263199329376, -0.04709354788064957, 0.03454873710870743, ...
corl_2023_rpWi4SYGXj
rpWi4SYGXj
corl
2,023
Grounding Complex Natural Language Commands for Temporal Tasks in Unseen Environments
Grounding navigational commands to linear temporal logic (LTL) leverages its unambiguous semantics for reasoning about long-horizon tasks and verifying the satisfaction of temporal constraints. Existing approaches require training data from the specific environment and landmarks that will be used in natural language to...
Jason Xinyu Liu;Ziyi Yang;Ifrah Idrees;Sam Liang;Benjamin Schornstein;Stefanie Tellex;Ankit Shah
Brown University;Brown University;Brown University;;Brown University;, Brown University;Brown University
Poster
main
language grounding;temporal reasoning;robot navigation;formal methods
https://github.com/h2r/Lang2LTL
https://openreview.net/forum?id=rpWi4SYGXj
46
Grounding Complex Natural Language Commands for Temporal Tasks in Unseen Environments Grounding navigational commands to linear temporal logic (LTL) leverages its unambiguous semantics for reasoning about long-horizon tasks and verifying the satisfaction of temporal constraints. Existing approaches require training dat...
[ -0.08407540619373322, 0.016276566311717033, 0.007732999045401812, -0.009391400963068008, -0.02208097279071808, -0.03225499019026756, -0.015521899797022343, -0.03518048673868179, -0.0414414219558239, 0.0033959979191422462, -0.008236110210418701, -0.00831530336290598, 0.010714396834373474, 0...
corl_2023_rvh0vkwKUM
rvh0vkwKUM
corl
2,023
Predicting Routine Object Usage for Proactive Robot Assistance
Proactivity in robot assistance refers to the robot's ability to anticipate user needs and perform assistive actions without explicit requests. This requires understanding user routines, predicting consistent activities, and actively seeking information to predict inconsistent behaviors. We propose SLaTe-PRO (Sequentia...
Maithili Patel;Aswin Gururaj Prakash;Sonia Chernova
Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology
Poster
main
Proactive Robot Assistance;User Routine Understanding;Interactive Clarification;Robot Learning
https://github.com/Maithili/SLaTe-PRO
https://openreview.net/forum?id=rvh0vkwKUM
10
Predicting Routine Object Usage for Proactive Robot Assistance Proactivity in robot assistance refers to the robot's ability to anticipate user needs and perform assistive actions without explicit requests. This requires understanding user routines, predicting consistent activities, and actively seeking information to ...
[ -0.05022036284208298, -0.01670622080564499, -0.02339055947959423, 0.024259615689516068, 0.027310557663440704, -0.006430092267692089, 0.012980373576283455, 0.042232438921928406, -0.023667916655540466, 0.026238104328513145, -0.04703998565673828, -0.017566032707691193, -0.025683388113975525, ...
corl_2023_rxlokRzNWRq
rxlokRzNWRq
corl
2,023
ManiCast: Collaborative Manipulation with Cost-Aware Human Forecasting
Seamless human-robot manipulation in close proximity relies on accurate forecasts of human motion. While there has been significant progress in learning forecast models at scale, when applied to manipulation tasks, these models accrue high errors at critical transition points leading to degradation in downstream planni...
Kushal Kedia;Prithwish Dan;Atiksh Bhardwaj;Sanjiban Choudhury
Cornell University;Department of Computer Science, Cornell University;Cornell University;Cornell University
Poster
main
Collaborative Manipulation;Forecasting;Model Predictive Control
https://openreview.net/forum?id=rxlokRzNWRq
5
ManiCast: Collaborative Manipulation with Cost-Aware Human Forecasting Seamless human-robot manipulation in close proximity relies on accurate forecasts of human motion. While there has been significant progress in learning forecast models at scale, when applied to manipulation tasks, these models accrue high errors at...
[ -0.020229989662766457, -0.004879480693489313, -0.04151894897222519, -0.005176175385713577, -0.013127584010362625, -0.0010989102302119136, 0.014159167185425758, 0.007549730129539967, 0.0030285646207630634, 0.0445132814347744, -0.026748135685920715, -0.038488104939460754, 0.01737259514629841, ...
corl_2023_sLhk0keeiseH
sLhk0keeiseH
corl
2,023
That Sounds Right: Auditory Self-Supervision for Dynamic Robot Manipulation
Learning to produce contact-rich, dynamic behaviors from raw sensory data has been a longstanding challenge in robotics. Prominent approaches primarily focus on using visual and tactile sensing. However, pure vision often fails to capture high-frequency interaction, while current tactile sensors can be too delicate for...
Abitha Thankaraj;Lerrel Pinto
;New York University
Poster
main
Dynamic manipulation;Self supervised learning;Audio
https://github.com/abitha-thankaraj/audio-robot-learning
https://openreview.net/forum?id=sLhk0keeiseH
12
That Sounds Right: Auditory Self-Supervision for Dynamic Robot Manipulation Learning to produce contact-rich, dynamic behaviors from raw sensory data has been a longstanding challenge in robotics. Prominent approaches primarily focus on using visual and tactile sensing. However, pure vision often fails to capture high-...
[ -0.03728483244776726, 0.011460656300187111, -0.026351969689130783, 0.031497955322265625, -0.0033717514015734196, 0.01419387198984623, 0.006394782103598118, 0.018972286954522133, -0.032855138182640076, 0.06981952488422394, -0.02478743903338909, -0.034438516944646835, -0.010518168099224567, ...
corl_2023_uJqxFjF1xWp
uJqxFjF1xWp
corl
2,023
BM2CP: Efficient Collaborative Perception with LiDAR-Camera Modalities
Collaborative perception enables agents to share complementary perceptual information with nearby agents. This can significantly benefit the perception performance and alleviate the issues of single-view perception, such as occlusion and sparsity. Most proposed approaches mainly focus on single modality (especially LiD...
Binyu Zhao;Wei ZHANG;Zhaonian Zou
Harbin Institute of Technology;Harbin Institute of Technology;Harbin Institute of Technology
Poster
main
Multi-Agent Perception;Multi-Modal Fusion;Vehicle-to-Everything (V2X) Application
https://openreview.net/forum?id=uJqxFjF1xWp
11
BM2CP: Efficient Collaborative Perception with LiDAR-Camera Modalities Collaborative perception enables agents to share complementary perceptual information with nearby agents. This can significantly benefit the perception performance and alleviate the issues of single-view perception, such as occlusion and sparsity. M...
[ -0.06210537627339363, -0.023965975269675255, -0.01866474561393261, 0.021536244079470634, -0.03055569715797901, 0.00033103922032751143, -0.0002526654861867428, 0.019566690549254417, 0.020192530006170273, 0.03057410567998886, -0.05032486468553543, -0.01864633895456791, 0.017284216359257698, ...
corl_2023_uo937r5eTE
uo937r5eTE
corl
2,023
Robot Parkour Learning
Parkour is a grand challenge for legged locomotion that requires robots to overcome various obstacles rapidly in complex environments. Existing methods can generate either diverse but blind locomotion skills or vision-based but specialized skills by using reference animal data or complex rewards. However, \textit{auton...
Ziwen Zhuang;Zipeng Fu;Jianren Wang;Christopher G Atkeson;Sören Schwertfeger;Chelsea Finn;Hang Zhao
ShanghaiTech University;Stanford University;Carnegie Mellon University;;;Google;Tsinghua University
Oral
main
Agile Locomotion;End-to-End Vision-Based Control;Sim-to-Real
https://github.com/ZiwenZhuang/parkour
https://openreview.net/forum?id=uo937r5eTE
195
Robot Parkour Learning Parkour is a grand challenge for legged locomotion that requires robots to overcome various obstacles rapidly in complex environments. Existing methods can generate either diverse but blind locomotion skills or vision-based but specialized skills by using reference animal data or complex rewards....
[ -0.05481158569455147, -0.007387728895992041, -0.005848233122378588, 0.033563774079084396, -0.0246319267898798, 0.0178082175552845, -0.011927622370421886, 0.0018977863946929574, 0.01859414577484131, -0.021987508982419968, -0.03949984535574913, -0.00733687449246645, -0.005145520903170109, 0....
corl_2023_veGdf4L4Xz
veGdf4L4Xz
corl
2,023
KITE: Keypoint-Conditioned Policies for Semantic Manipulation
While natural language offers a convenient shared interface for humans and robots, enabling robots to interpret and follow language commands remains a longstanding challenge in manipulation. A crucial step to realizing a performant instruction-following robot is achieving semantic manipulation – where a robot interpret...
Priya Sundaresan;Suneel Belkhale;Dorsa Sadigh;Jeannette Bohg
Stanford University;Stanford University;Stanford University;Stanford University
Poster
main
Semantic Manipulation;Language Grounding;Keypoint Perception
http://tinyurl.com/kite-site
https://openreview.net/forum?id=veGdf4L4Xz
25
KITE: Keypoint-Conditioned Policies for Semantic Manipulation While natural language offers a convenient shared interface for humans and robots, enabling robots to interpret and follow language commands remains a longstanding challenge in manipulation. A crucial step to realizing a performant instruction-following robo...
[ -0.017044078558683395, 0.0045544845052063465, -0.016138264909386635, 0.01009334810078144, -0.02201680839061737, -0.02255290187895298, 0.006289856508374214, -0.018874190747737885, -0.020316097885370255, 0.01468711532652378, 0.009242992848157883, -0.03347811847925186, -0.00009741247777128592, ...
corl_2023_vsEWu6mMUhB
vsEWu6mMUhB
corl
2,023
Semantic Mechanical Search with Large Vision and Language Models
Moving objects to find a fully-occluded target object, known as mechanical search, is a challenging problem in robotics. As objects are often organized semantically, we conjecture that semantic information about object relationships can facilitate mechanical search and reduce search time. Large pretrained vision and la...
Satvik Sharma;Huang Huang;Kaushik Shivakumar;Lawrence Yunliang Chen;Ryan Hoque;brian ichter;Ken Goldberg
University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Google;University of California, Berkeley
Poster
main
Vision and Language Models in Robotics;Mechanical Search;Object search
https://openreview.net/forum?id=vsEWu6mMUhB
11
Semantic Mechanical Search with Large Vision and Language Models Moving objects to find a fully-occluded target object, known as mechanical search, is a challenging problem in robotics. As objects are often organized semantically, we conjecture that semantic information about object relationships can facilitate mechani...
[ -0.06849350780248642, -0.017104854807257652, -0.0030028419569134712, -0.008589470759034157, -0.013243066146969795, -0.005542639642953873, -0.011835411190986633, -0.03837710991501808, 0.011613150127232075, 0.018632901832461357, -0.003984495997428894, 0.00959427747875452, 0.019892381504178047,...
corl_2023_w5ONmpgnfG
w5ONmpgnfG
corl
2,023
One-Shot Imitation Learning: A Pose Estimation Perspective
In this paper, we study imitation learning under the challenging setting of: (1) only a single demonstration, (2) no further data collection, and (3) no prior task or object knowledge. We show how, with these constraints, imitation learning can be formulated as a combination of trajectory transfer and unseen object pos...
Pietro Vitiello;Kamil Dreczkowski;Edward Johns
Imperial College London;Imperial College London;Imperial College London
Poster
main
One-Shot Imitation Learning;Unseen Object Pose Estimation;Robot Manipulation
https://openreview.net/forum?id=w5ONmpgnfG
23
One-Shot Imitation Learning: A Pose Estimation Perspective In this paper, we study imitation learning under the challenging setting of: (1) only a single demonstration, (2) no further data collection, and (3) no prior task or object knowledge. We show how, with these constraints, imitation learning can be formulated as...
[ -0.04260310158133507, -0.0011755266459658742, -0.01636113040149212, 0.01397799700498581, 0.012502286583185196, 0.013281388208270073, -0.01242895983159542, 0.02989916317164898, 0.028304297477006912, -0.012713102623820305, -0.041796501725912094, -0.032098978757858276, 0.014161314815282822, 0...
corl_2023_wH23nZpVTF6
wH23nZpVTF6
corl
2,023
DEFT: Dexterous Fine-Tuning for Hand Policies
Dexterity is often seen as a cornerstone of complex manipulation. Humans are able to perform a host of skills with their hands, from making food to operating tools. In this paper, we investigate these challenges, especially in the case of soft, deformable objects as well as complex, relatively long-horizon tasks. Alth...
Aditya Kannan;Kenneth Shaw;Shikhar Bahl;Pragna Mannam;Deepak Pathak
School of Computer Science, Carnegie Mellon University;Carnegie Mellon University;Meta Facebook;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Dexterous Manipulation;Reinforcement Learning;Learning from Videos
https://openreview.net/forum?id=wH23nZpVTF6
-1
DEFT: Dexterous Fine-Tuning for Hand Policies Dexterity is often seen as a cornerstone of complex manipulation. Humans are able to perform a host of skills with their hands, from making food to operating tools. In this paper, we investigate these challenges, especially in the case of soft, deformable objects as well a...
[ -0.0014508464373648167, -0.023457305505871773, -0.03945177420973778, -0.019744638353586197, -0.023119788616895676, -0.013444354757666588, 0.0033915587700903416, 0.0108473626896739, 0.014672534540295601, 0.03063887730240822, -0.035814110189676285, -0.007955045439302921, 0.003904863027855754, ...
corl_2023_wMpOMO0Ss7a
wMpOMO0Ss7a
corl
2,023
SayPlan: Grounding Large Language Models using 3D Scene Graphs for Scalable Robot Task Planning
Large language models (LLMs) have demonstrated impressive results in developing generalist planning agents for diverse tasks. However, grounding these plans in expansive, multi-floor, and multi-room environments presents a significant challenge for robotics. We introduce SayPlan, a scalable approach to LLM-based, large...
Krishan Rana;Jesse Haviland;Sourav Garg;Jad Abou-Chakra;Ian Reid;Niko Suenderhauf
Queensland University of Technology;Queensland University of Technology;Queensland University of Technology;Queensland University of Technology;University of Adelaide;Queensland University of Technology
Oral
main
robot task planning;large language models;semantic search;LLM-based planning;3D scene graphs
https://openreview.net/forum?id=wMpOMO0Ss7a
315
SayPlan: Grounding Large Language Models using 3D Scene Graphs for Scalable Robot Task Planning Large language models (LLMs) have demonstrated impressive results in developing generalist planning agents for diverse tasks. However, grounding these plans in expansive, multi-floor, and multi-room environments presents a s...
[ -0.056175343692302704, -0.032150089740753174, -0.0053813261911273, -0.019484903663396835, 0.00154408672824502, -0.000883484142832458, 0.0009047383209690452, -0.010542068630456924, -0.009889507666230202, 0.019889308139681816, 0.005762752145528793, 0.026010507717728615, 0.023179681971669197, ...
corl_2023_xJ7XL5Wt8iN
xJ7XL5Wt8iN
corl
2,023
CLUE: Calibrated Latent Guidance for Offline Reinforcement Learning
Offline reinforcement learning (RL) aims to learn an optimal policy from pre-collected and labeled datasets, which eliminates the time-consuming data collection in online RL. However, offline RL still bears a large burden of specifying/handcrafting extrinsic rewards for each transition in the offline data. As a remedy ...
Jinxin Liu;Lipeng Zu;Li He;Donglin Wang
;;Westlake University;Westlake University
Poster
main
Offline Reinforcement Learning;Intrinsic Rewards;Learning Skills
https://openreview.net/forum?id=xJ7XL5Wt8iN
9
CLUE: Calibrated Latent Guidance for Offline Reinforcement Learning Offline reinforcement learning (RL) aims to learn an optimal policy from pre-collected and labeled datasets, which eliminates the time-consuming data collection in online RL. However, offline RL still bears a large burden of specifying/handcrafting ext...
[ -0.09410450607538223, 0.019499467685818672, -0.02382439747452736, 0.00638486398383975, -0.007689799647778273, -0.01843687705695629, 0.008533347398042679, -0.00666449312120676, 0.0037656715139746666, 0.014335650950670242, -0.022910941392183304, -0.024327728897333145, -0.023451559245586395, ...
corl_2023_xQx1O7WXSA
xQx1O7WXSA
corl
2,023
Expansive Latent Planning for Sparse Reward Offline Reinforcement Learning
Sampling-based motion planning algorithms excel at searching global solution paths in geometrically complex settings. However, classical approaches, such as RRT, are difficult to scale beyond low-dimensional search spaces and rely on privileged knowledge e.g. about collision detection and underlying state distances. In...
Robert Gieselmann;Florian T. Pokorny
KTH Royal Institute of Technology, Stockholm, Sweden;
Oral
main
model-based reinforcement learning;planning;robot manipulation
https://openreview.net/forum?id=xQx1O7WXSA
1
Expansive Latent Planning for Sparse Reward Offline Reinforcement Learning Sampling-based motion planning algorithms excel at searching global solution paths in geometrically complex settings. However, classical approaches, such as RRT, are difficult to scale beyond low-dimensional search spaces and rely on privileged ...
[ -0.05555596575140953, -0.0044251405633986, 0.005563963670283556, 0.039194103330373764, -0.01720784790813923, 0.02731315791606903, 0.04518105834722519, 0.009528926573693752, -0.0018232787260785699, 0.005071248393505812, -0.017458852380514145, -0.0023043733090162277, -0.025472447276115417, 0...
corl_2023_xgrZkRHliXR
xgrZkRHliXR
corl
2,023
Learning to Design and Use Tools for Robotic Manipulation
When limited by their own morphologies, humans and some species of animals have the remarkable ability to use objects from the environment toward accomplishing otherwise impossible tasks. Robots might similarly unlock a range of additional capabilities through tool use. Recent techniques for jointly optimizing morpholo...
Ziang Liu;Stephen Tian;Michelle Guo;Karen Liu;Jiajun Wu
;Stanford University;Computer Science Department, Stanford University;Computer Science Department, Stanford University;Stanford University
Poster
main
tool use;manipulation;design
https://openreview.net/forum?id=xgrZkRHliXR
4
Learning to Design and Use Tools for Robotic Manipulation When limited by their own morphologies, humans and some species of animals have the remarkable ability to use objects from the environment toward accomplishing otherwise impossible tasks. Robots might similarly unlock a range of additional capabilities through t...
[ -0.05194586515426636, -0.02691841311752796, -0.024341514334082603, 0.022728633135557175, -0.03374071046710014, -0.01200390700250864, -0.019039401784539223, 0.022895483300089836, 0.009046959690749645, 0.01461788546293974, -0.019076479598879814, -0.020300043746829033, 0.0219685398042202, -0....
corl_2023_yGkqN4hqrJ
yGkqN4hqrJ
corl
2,023
Fine-Tuning Generative Models as an Inference Method for Robotic Tasks
Adaptable models could greatly benefit robotic agents operating in the real world, allowing them to deal with novel and varying conditions. While approaches such as Bayesian inference are well-studied frameworks for adapting models to evidence, we build on recent advances in deep generative models which have greatly af...
Orr Krupnik;Elisei Shafer;Tom Jurgenson;Aviv Tamar
Technion - Israel Institute of Technology, Technion - Israel Institute of Technology;;Technion;Technion, Technion
Poster
main
robotic learning;fine-tuning;generative models
https://github.com/orrkrup/mace/
https://openreview.net/forum?id=yGkqN4hqrJ
3
Fine-Tuning Generative Models as an Inference Method for Robotic Tasks Adaptable models could greatly benefit robotic agents operating in the real world, allowing them to deal with novel and varying conditions. While approaches such as Bayesian inference are well-studied frameworks for adapting models to evidence, we b...
[ 0.0004487091791816056, -0.02734553813934326, -0.02646755427122116, 0.017971232533454895, -0.01717555895447731, -0.08304263651371002, 0.03665582463145256, 0.01354472991079092, 0.01913273148238659, 0.020870408043265343, -0.05110597610473633, -0.008427729830145836, -0.009538928978145123, 0.02...
corl_2023_yHlUVHWnBN
yHlUVHWnBN
corl
2,023
SCONE: A Food Scooping Robot Learning Framework with Active Perception
Effectively scooping food items poses a substantial challenge for current robotic systems, due to the intricate states and diverse physical properties of food. To address this challenge, we believe in the importance of encoding food items into meaningful representations for effective food scooping. However, the distinc...
Yen-Ling Tai;Yu Chien Chiu;Yu-Wei Chao;Yi-Ting Chen
National Yang Ming Chiao Tung University;National Yang Ming Chiao Tung University;NVIDIA;National Yang Ming Chiao Tung University
Poster
main
Food Manipulation;Robot Scooping;Active Perception
https://openreview.net/forum?id=yHlUVHWnBN
12
SCONE: A Food Scooping Robot Learning Framework with Active Perception Effectively scooping food items poses a substantial challenge for current robotic systems, due to the intricate states and diverse physical properties of food. To address this challenge, we believe in the importance of encoding food items into meani...
[ -0.009045287035405636, -0.0008520922274328768, -0.010262561030685902, 0.000896569574251771, -0.008188513107597828, -0.00507041672244668, -0.00703678373247385, 0.024869857355952263, 0.006507737562060356, 0.02082476019859314, 0.02672385983169079, -0.004967416636645794, -0.025693858042359352, ...
corl_2023_ycy47ZX0Oc
ycy47ZX0Oc
corl
2,023
Leveraging 3D Reconstruction for Mechanical Search on Cluttered Shelves
Finding and grasping a target object on a cluttered shelf, especially when the target is occluded by other unknown objects and initially invisible, remains a significant challenge in robotic manipulation. While there have been advances in finding the target object by rearranging surrounding objects using specialized to...
Seungyeon Kim;Young Hun Kim;Yonghyeon Lee;Frank C. Park
Seoul National University;Seoul National University;Seoul National University;Seoul National University
Poster
main
Mechanical search;Object rearrangement;Prehensile and Non-prehensile manipulation
https://github.com/seungyeon-k/Search-for-Grasp-public
https://openreview.net/forum?id=ycy47ZX0Oc
4
Leveraging 3D Reconstruction for Mechanical Search on Cluttered Shelves Finding and grasping a target object on a cluttered shelf, especially when the target is occluded by other unknown objects and initially invisible, remains a significant challenge in robotic manipulation. While there have been advances in finding t...
[ -0.034111008048057556, -0.047812800854444504, -0.0304703488945961, -0.007617587223649025, -0.007151295896619558, -0.006873314268887043, 0.03448762744665146, -0.03502565622329712, -0.007765545044094324, 0.009863856248557568, -0.01727968268096447, 0.011325500905513763, 0.002591504016891122, ...
corl_2023_yobahDU4HPP
yobahDU4HPP
corl
2,023
Learning Realistic Traffic Agents in Closed-loop
Realistic traffic simulation is crucial for developing self-driving software in a safe and scalable manner prior to real-world deployment. Typically, imitation learning (IL) is used to learn human-like traffic agents directly from real-world observations collected offline, but without explicit specification of traffic ...
Chris Zhang;James Tu;Lunjun Zhang;Kelvin Wong;Simon Suo;Raquel Urtasun
Department of Computer Science, University of Toronto;Waabi Innovation;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto
Poster
main
Traffic simulation;Imitation learning;Reinforcement learning
https://openreview.net/forum?id=yobahDU4HPP
19
Learning Realistic Traffic Agents in Closed-loop Realistic traffic simulation is crucial for developing self-driving software in a safe and scalable manner prior to real-world deployment. Typically, imitation learning (IL) is used to learn human-like traffic agents directly from real-world observations collected offline...
[ -0.09636995941400528, -0.007980176247656345, 0.04004363343119621, 0.028089484199881554, -0.035162508487701416, 0.006962508428841829, -0.005502776242792606, 0.019432397559285164, -0.027905290946364403, 0.0265975184738636, -0.06741476058959961, -0.042548660188913345, -0.008795231580734253, 0...
corl_2023_z3D__-nc9y
z3D__-nc9y
corl
2,023
Autonomous Robotic Reinforcement Learning with Asynchronous Human Feedback
Ideally, we would place a robot in a real-world environment and leave it there improving on its own by gathering more experience autonomously. However, algorithms for autonomous robotic learning have been challenging to realize in the real world. While this has often been attributed to the challenge of sample complexit...
Max Balsells I Pamies;Marcel Torne Villasevil;Zihan Wang;Samedh Desai;Pulkit Agrawal;Abhishek Gupta
Universidad Politécnica de Cataluna;Harvard University, Harvard University;University of Washington;University of Washington;Massachusetts Institute of Technology;University of Washington
Poster
main
reset-free reinforcement learning;learning from human feedback
https://github.com/guided-exploration-autonomous-rl/gear-code/tree/main
https://openreview.net/forum?id=z3D__-nc9y
6
Autonomous Robotic Reinforcement Learning with Asynchronous Human Feedback Ideally, we would place a robot in a real-world environment and leave it there improving on its own by gathering more experience autonomously. However, algorithms for autonomous robotic learning have been challenging to realize in the real world...
[ -0.08994963020086288, 0.0078982412815094, -0.015132919885218143, -0.0384129174053669, -0.03638536483049393, -0.022837622091174126, 0.026468785479664803, 0.021694818511605263, 0.0037256290670484304, 0.0052578141912817955, -0.027021754533052444, -0.015280378051102161, -0.04099343717098236, 0...
corl_2023_zUiH8UUYDo
zUiH8UUYDo
corl
2,023
Scalable Deep Kernel Gaussian Process for Vehicle Dynamics in Autonomous Racing
Autonomous racing presents a challenging environment for testing the limits of autonomous vehicle technology. Accurately modeling the vehicle dynamics (with all forces and tires) is critical for high-speed racing, but it remains a difficult task and requires an intricate balance between run-time computational demands a...
Jingyun Ning;Madhur Behl
University of Virginia, Charlottesville;University of Virginia
Poster
main
Gaussian Process;Vehicle Dynamics;Autonomous Vehicle;Deep Kernel Learning
https://openreview.net/forum?id=zUiH8UUYDo
6
Scalable Deep Kernel Gaussian Process for Vehicle Dynamics in Autonomous Racing Autonomous racing presents a challenging environment for testing the limits of autonomous vehicle technology. Accurately modeling the vehicle dynamics (with all forces and tires) is critical for high-speed racing, but it remains a difficult...
[ -0.0872892439365387, -0.039261896163225174, -0.009596338495612144, 0.003108994336798787, -0.03904275968670845, -0.018489614129066467, 0.03033209778368473, -0.033564355224370956, -0.009705905802547932, 0.03871405869722366, -0.028925973922014236, 0.002839639550074935, 0.01141334231942892, 0....
corl_2023_zvl2LuLTtgr
zvl2LuLTtgr
corl
2,023
What Went Wrong? Closing the Sim-to-Real Gap via Differentiable Causal Discovery
Training control policies in simulation is more appealing than on real robots directly, as it allows for exploring diverse states in an efficient manner. Yet, robot simulators inevitably exhibit disparities from the real-world \rebut{dynamics}, yielding inaccuracies that manifest as the dynamical simulation-to-reality ...
Peide Huang;Xilun Zhang;Ziang Cao;Shiqi Liu;Mengdi Xu;Wenhao Ding;Jonathan Francis;Bingqing Chen;Ding Zhao
Carnegie Mellon University;Carnegie Mellon University;;;Carnegie Mellon University;Carnegie Mellon University;;Bosch;Carnegie Mellon University
Poster
main
sim-to-real gap;reinforcement learning;causal discovery
https://openreview.net/forum?id=zvl2LuLTtgr
33
What Went Wrong? Closing the Sim-to-Real Gap via Differentiable Causal Discovery Training control policies in simulation is more appealing than on real robots directly, as it allows for exploring diverse states in an efficient manner. Yet, robot simulators inevitably exhibit disparities from the real-world \rebut{dynam...
[ -0.03500966355204582, -0.03925269469618797, -0.015878649428486824, 0.0061355698853731155, -0.035177890211343765, -0.03558911010622978, -0.005757061764597893, -0.0048271468840539455, -0.004976680967956781, 0.06287907809019089, -0.021551597863435745, 0.015271167270839214, 0.0017138005932793021...
corl_2024_0M7JiV1GFN
0M7JiV1GFN
corl
2,024
Provably Safe Online Multi-Agent Navigation in Unknown Environments
Control Barrier Functions (CBFs) provide safety guarantees for multi-agent navigation. However, traditional approaches require full knowledge of the environment (e.g., obstacle positions and shapes) to formulate CBFs and hence, are not applicable in unknown environments. This paper overcomes this issue by proposing an ...
Zhan Gao;Guang Yang;Jasmine Bayrooti;Amanda Prorok
;University of Cambridge;University of Cambridge;
Poster
main
Decentralized Multi-Agent Navigation;Unknown Environment;Support Vector Machine;Graph Attention Learning;Control Barrier Function
https://openreview.net/forum?id=0M7JiV1GFN
1
Provably Safe Online Multi-Agent Navigation in Unknown Environments Control Barrier Functions (CBFs) provide safety guarantees for multi-agent navigation. However, traditional approaches require full knowledge of the environment (e.g., obstacle positions and shapes) to formulate CBFs and hence, are not applicable in un...
[ -0.03051052615046501, -0.00923982448875904, -0.06441318243741989, 0.017407959327101707, -0.0018556531285867095, 0.006611856631934643, 0.028218042105436325, 0.0016226771986111999, 0.013559198006987572, 0.042122043669223785, -0.03714567795395851, -0.0038347835652530193, 0.0032896199263632298, ...