ICRA 2024 Workshop

Back to the Future: Robot Learning Going Probabilistic
Date: May 13th, 2024 | Room: F202, Annex Hall
The Hybrid events will be accessible to all registered participants from Infovaya.
Link to: Virtual Poster Session Gather Town (10-11AM JST)

About

Probabilistic robotics, a vibrant field that has gained enormous popularity since its inception, provides a compelling paradigm for autonomous robots to contend with the complex real world. Probabilistic representations yield beneficial properties for trustworthy learning-enabled robots deployed in the real world, e.g., uncertainty estimation, ways to elegantly handle incomplete data and the unifying perspective on perception, control and learning. On the other hand, recent advances in deep learning have dramatically improved the suitability and performance of robot learning, e.g., large language models (LLMs), visual foundational models, and Neural Radiance Fields (NeRFs), to name a few. Though there have been advances in pursuing the probabilistic extension of these concepts in recent years, many core challenges associated with real-world deployment remain unsolved.

In light of the above, this workshop aims to provide a forum to bring together robotic and machine learning researchers as well as industry experts with experience in developing probabilistic methods that dovetail with robot learning. In order to facilitate breakthrough research in these areas, the discussions will be centred on past achievements, current requirements, urgent challenges and future directions to enable promising applications.


Speakers

Andreas Krause
Ayoung Kim
Peter Karkus
Andreas Krause Ayoung Kim Peter Karkus
ETH Zürich & LatticeFlow (Switzerland) Seoul National University (Korea) NVIDIA (USA)
Talk: Safe and Efficient Exploration in Bayesian Model-based Reinforcement Learning Talk: Deep learning with a graph SLAM for robot manipulation Talk: End-to-end yet Modular Probabilistic Robot Learning
Niko Sünderhauf
Juho Lee
Sharon Yixuan Li
Masashi Sugiyama
Niko Sünderhauf Juho Lee Sharon Yixuan Li Masashi Sugiyama
Queensland University of Technology (Australia) KAIST (Korea) University of Wisconsin (USA) RIKEN and University of Tokyo (Japan)
Talk: Predictive Uncertainty Neural Radiance Fields for Robotics Talk: Scaling Bayesian deep learning for fast and accurate inference Talk: How to Detect Out-of-Distribution Data in the Wild? Challenges, Progress, and Opportunities Talk: Learning under Sequential Distribution Shifts

Accepted Papers

Title Authors
3D Diffuser Actor: Policy Diffusion with 3D Scene Representations Tsung-Wei Ke, Nikolaos Gkanatsios, and Katerina Fragkiadaki
Generative Factor Chaining: Coordinated Manipulation with Diffusion-based Factor Graph Utkarsh Aashu Mishra, Yongxin Chen, and Danfei Xu
uPLAM: Robust Panoptic Localization and Mapping Leveraging Perception Uncertainties Kshitij Sirohi, Daniel Büscher, and Wolfram Burgard
Latent Space Exploration and Trajectory Space Update in Temporally-Correlated Episodic Reinforcement Learning Ge Li, Hongyi Zhou, Dominik Roth, Serge Thilges, Fabian Otto, Rudolf Lioutikov, and Gerhard Neumann
Safe Offline Reinforcement Learning using Trajectory-Level Diffusion Models Ralf Römer, Lukas Brunke, Martin Schuck, and Angela P. Schoellig
Risk-Calibrated Human-Robot Interaction via Set-Valued Intent Prediction Justin Lidard, Hang Pham, Ariel Bachman, Bryan Boateng, and Anirudha Majumdar
PRIMP: PRobabilistically-Informed Motion Primitives for Efficient Affordance Learning from Demonstration Sipu Ruan, Weixiao Liu, Xiaoli Wang, Xin Meng, and Gregory S Chirikjian
Revisiting Semantic Class Uncertainties for Robust Visual Place Recognition Alex Junho Lee and Dong jin Hyun
Evaluating Uncertainty-based Failure Detection for Closed-Loop LLM Planners Zhi Zheng, Qian Feng, hang li, Alois Knoll, and Jianxiang Feng
Zero-shot Safety Prediction for Autonomous Robots with Foundation World Models Zhenjiang Mao, Siqi Dai, Yuang Geng, and Ivan Ruchkin
Enabling Stateful Behaviors for Diffusion-based Policy Learning Xiao Liu, Fabian C Weigend, Yifan Zhou, and Heni Ben Amor
Learning Emergent Gaits with Decentralized Phase Oscillators: on the role of Observations, Rewards, and Feedback Jenny Zhang, Steve Heim, Se Hwan Jeon, and Sang bae Kim
LiRA: Light-Robust Adversary for Model-based Reinforcement Learning Taisuke Kobayashi
Stochastic Implicit Neural Signed Distance Functions for Safe Motion Planning under Sensing Uncertainty Carlos Quintero-Pena, Wil Thomason, Zachary Kingston, Anastasios Kyrillidis, and Lydia E Kavraki
Spatial Reasoning with Open Set Vocabulary Object Detectors for Robot Perception Negar Nejatishahidin and Jana Kosecka
EDMP: Ensemble-of-costs-guided Diffusion for Motion Planning Kallol Saha, Vishal Mandadi, Jayaram Reddy, Ajit Srikanth, Aditya Agarwal, Bipasha Sen, Arun Kumar Singh, and Madhava Krishna

Schedule

Time Speaker Session/Title
08:45-09:00 Rudolph Triebel Welcome message from organisers
09:00-09:30 Andreas Krause Safe and Efficient Exploration in Bayesian Model-based Reinforcement Learning
09:30-10:00 Juho Lee Scaling Bayesian deep learning for fast and accurate inference
10:00-10:30   Interactive poster session
10:30-11:00   Coffee break
11:00-11:30 Masashi Sugiyama Learning under Sequential Distribution Shifts
11:30-12:00 Sharon Li How to Detect Out-of-Distribution Data in the Wild? Challenges, Progress, and Opportunities
12:00-13:00   Lunch break
13:00-13:30 Niko Sünderhauf Predictive Uncertainty Neural Radiance Fields for Robotics
13:30-14:00 Peter Karkus From Differentiable Algorithms to Robot Foundation Models
14:00-14:20 Utkarsh Aashu Mishra Generative Factor Chaining: Coordinated Manipulation with
14:20-14:40 Justin Lidard Risk-Calibrated Human-Robot Interaction via Set-Valued Intent Prediction
14:40-15:00 Kshitij Sirohi uPLAM: Robust Panoptic Localization and Mapping Leveraging Perception Uncertainties
Diffusion-based Factor Graph    
15:00-15:30   Coffee break
15:30-16:00 Ayoung Kim Deep learning with a graph SLAM for robot manipulation
16:00-17:00   Panel discussion

Committees

Organizing Committee

Anthony Opipari
Jana Pavlasek
Jianxiang Feng
Jongseok Lee
Yizhe Wu
Rudolph Triebel
Anthony Opipari Jana Pavlasek Jianxiang Feng Jongseok Lee Yizhe Wu Rudolph Triebel
University of Michigan (USA) University of Michigan (USA) TU Munich & Agile Robots (Germany) DLR & KIT (Germany) University of Oxford (United Kingdom) DLR & Karlsruhe Institute of Technology (KIT) (Germany)

Scientific Committee

Janis Postels
Matthias Humt
Thomas Power
Chad Jenkins
Tucker Hermans
Fabio Ramos
Janis Postels Matthias Humt Thomas Power Chad Jenkins Tucker Hermans Fabio Ramos
ETH Zurich (Switzerland) DLR & TU Munich (Germany) University of Michigan (USA) University of Michigan (USA) University of Utah & NVIDIA (USA) University of Sydney & NVIDIA
(Australia & USA)

Sponsors

This workshop and its best paper awards are generously sponsored by Agile Robots AG and Amazon Consumer Robotics


This workshop is proud to be endorsed by the IEEE RAS Technical Committees for Computer and Robot Vision and Autonomous Ground Vehicles and Intelligent Transportation Systems

Contact

Should you have any questions, please do not hesitate to contact the organizing committee