Scaling Robot Learning
ICRA 2022 Workshop | May 27, 2022
Description
With an increased adoption of thousands of robots in various industries comes the question of how to better deploy physical robots “at scale”. In this workshop, we are specifically interested in the challenges connected to “scaling” machine learning for robotic applications. This includes topics on both the algorithmic advancements (self-supervised learning, lifelong learning, multi-task learning, etc), and the systems side (data-acquisition at scale, distributed training, real-world challenges in deploying and maintaining a fleet of thousands of robots). The goal of this workshop is to bring together experts from different communities, industry and academia, to discuss challenges and explore new directions towards scaling robot learning.
The workshop will consist of invited talks, spotlight presentations, a poster session and interdisciplinary panel discussions. We aim to improve the communication across a diverse set of scientists who are at various stages of their careers. Instead of the common trade-offs between attracting a wider audience with well-known speakers and enabling early-stage researchers to voice their opinion, we encourage each of our senior presenters to share their presentations with a PhD student or postdoc from their lab. We also ask all our presenters - invited and contributed - to add a “dirty laundry” slide, describing limitations of their work. We expect this will aid further discussion in poster and panel sessions in addition to helping junior researchers avoid similar roadblocks along their path.
The focus topics for our workshop include, but are not restricted to:
Self-supervised learning for robotics, robot vision, reinforcement learning….
Lifelong learning
Multi-task learning
Data acquisition at scale
Multi robot systems / learning from fleets
Large scale robotics applications and systems
Scaling task specification via language/video conditioned policies
Thanks everybody for participating!
Invited Speakers
Dieter Fox
U Washington / Nvidia
Jeannette Bohg
Standford
Kendra Byrne
Robotics @ Google AI
Stephanie Tellex
Brown University
Yuning Chai
Cruise
Sean Wilson
Georgia Tech
Andy Marchese
Amazon Robotics
Thanks for Participating!
Live Streams
Poster Sessions Gather.Town: https://app.gather.town/app/P6V9fyCw1AxuUGWu/Scaling%20Robot%20Learning Youtube Recordings: https://www.youtube.com/watch?v=EV0gSmmUN5c&list=PLkWItR7T--mfVd4nQy20_WxFmN17xzeOi&index=1
Zoom: https://zoom.us/j/4874613097?pwd=UWdrMHVNb1dQbEJGWU54cG40N2kwZz09
Submit your questions for the panel discussions here: https://forms.gle/rfm7fnDpSR31FsmY6
Schedule
Friday, 27th of May, Easter Time
Room: 108B
8:30 - 8:35 Welcome organizers
8:35 - 9:00 Invited Talk 1 (Dieter Fox)
9:00 - 9:25 Invited Talk 2 (Andy Marchese)
9:25 - 9:45 Spotlight Sessions
Equivariant Reinforcement Learning for Robotic Manipulation
Learning Perceptual Concepts by Bootstrapping from Human Queries
9.45 - 10:15 Contributed Papers Discussions
10:15 - 11:00 Coffee Break
11:00 - 11:25 Invited Talk 3 (Jeannette Bohg)
11:25 - 12:10 QA + Panel 1 (Dieter Fox, Andy Marchese, Jeannette Bohg)
13:00 - 13:25 Invited Talk 5 (Kendra Byrne)
13:25 - 13:50 Invited Talk 6 (Stephanie Tellex)
13:50 - 14:45 Spotlight Sessions:
Jump-Start Reinforcement Learning
Meta-Learning Regrasping Strategies for Physical-Agnostic Objects
Sim-to-Real Transfer for High-Speed Quadrotor Flight
Sample Efficient Grasp Learning Using Equivariant Models
Equivariant Transporter Network
R3M: A Universal Visual Representation for Robot Manipulation
SORNet: Spatial Object-Centric Representations for Sequential Manipulation
14:45 - 15:15 Contributed Papers Discussions
15:15 - 16:00 Coffee Break
16:00 - 16:25 Invited Talk 7 (Yuning Chai)
16:25 - 16:50 Invited Talk 8 (Sean Wilson)
16:50 - 17:35 QA + Panel 2 (Kendra Byrne, Stephanie Tellex, Yuning Chai, Sean Wilson)
17:35 - 17:45 Award & Concluding remarks
Accepted Papers
Evaluating Vision Transformer Methods for Deep Reinforcement Learning from Pixels [paper]
Bridging the Gap between Events and Frames through Unsupervised Domain Adaptation [paper]
Meta-Learning Regrasping Strategies for Physical-Agnostic Objects [paper]
T3VIP: Transformation-based 3D Video Prediction [paper]
Learning to Plan with Optimistic Action Models [paper]
An Object-Oriented Approach for Generating Low-Fidelity Environments for Curriculum Schema Transfer [paper]
Visually Adaptive Geometric Navigation [paper]
Embodied Active Domain Adaptation for Semantic Segmentation via Informative Path Planning [paper]
Equivariant Reinforcement Learning for Robotic Manipulation [paper]
Reconstructing occluded Elevation Information in Terrain Maps with Self-supervised Learning [paper]
Equivariant Transporter Network [paper]
Sim-to-Real Transfer for High-Speed Quadrotor Flight [paper]
Sample Efficient Grasp Learning Using Equivariant Models [paper]
Robot Learning of Mobile Manipulation with Reachability Behavior Priors [paper]
R3M: A Universal Visual Representation for Robot Manipulation [paper]
Learning Perceptual Concepts by Bootstrapping from Human Queries [paper]
Jump-Start Reinforcement Learning [paper]
SORNet: Spatial Object-Centric Representations for Sequential Manipulation [paper]
Minimum-time Planning for Aerial Robots in Cluttered Environments [paper]
Pedestrian Motion Pattern Prediction from Traversability Maps [paper]
Best Paper Award
R3M: A Universal Visual Representation for Robot Manipulation [paper]
Poster Instructions
For printing physical posters, please refer to the ICRA format guidelines https://www.icra2022.org/contribute/person-and-virtual-presentation-instructions
To accommodate multiple timezones in this hybrid workshop format authors should, if feasible, try and attend both sheduled sessions to present their poster. If not possible, please present at the session that best works for you.
Onsite authors should bring their webcam-enabled personal laptops to connect to gather.town for a hybrid poster session.
Call for Contributions
We encourage participants to submit their research in the form of a single PDF. Submissions may be up to 4 pages in length, including figures, excluding references and any supplementary material. Please use the ICRA conference template. Accepted papers will be presented in a poster session and selected papers as spotlight talks. All submitted contributions will go through a single blind review process. The contributed papers will be made available on the workshop’s website. However, this does not constitute an archival publication and no formal workshop proceedings will be made available, meaning contributors are free to publish their work in archival journals or conference.
Google is sponsoring a Best Paper Award of $1,000 to be shared by all authors of the awarded paper.
Submission Website: https://cmt3.research.microsoft.com/ICRAWSRL2022
LaTeX Template: http://ras.papercept.net/conferences/support/tex.php
Important Dates
April 11 18, 2022 Submission Deadline (AoE time)
April 25, 2022 Author Notification
May 12, 2022 Camera Ready Submission
May 27, 2022 Workshop
Organizers
TU Nuremberg
Facebook AI Research
DeepMind
INRIA
University of Freiburg
Google Research
For further information please contact us at icra22-srl(at)googlegroups.com