1st Workshop on 3D Visual Representations for Robot Manipulation.

Location: ICRA 2024, Yokohama, Japan
Date: May 17th, 2024 Time: 9:00am - 5:00pm

About

In order for robots to be capable of manipulating a wide range of novel objects in unstructured environments, they must have the ability to perceive and reason about 3D geometry. By doing so, robots will then be able to use its 3D visual information to better inform actions that it can execute in the 3D world. However, there are a number of challenges we must overcome. These challenges include: how to acquire and represent 3D information, and how to learn 3D representations which are useful for manipulation agents. Ultimately, we hope this workshop will help the community make progress towards understanding how we can build more generalist manipulation agents.

Important Dates

Call for Workshop Papers

We invite submissions of papers on the topic of 3D visual representations for robot manipulation. Topics of interest include, but are not limited to:

• Acquiring 3D geometric information
• Representing raw 3D geometry
• Learning on top of 3D geometric inputs
• Building 3D representations which are useful for manipulation agents
• Connecting 3D visual representations to action spaces
• Utility of foundation models / LLMs for visual manipulation tasks

Please see the call for papers for information on format and submission.

Invited Speakers

Jeannette Bohg
Stanford University

Jen Jen Chung
University of Queensland

Pete Florence
Google DeepMind

Katerina Fragkiadaki
Carnegie Mellon University

Stephen James
Dyson Robotics

Hao Su
University of California, San Diego

He Wang
Peking University

Yu Xiang
University of Texas, Dallas

Tentative Schedule

Organizers

Ben Eisner
Carnegie Mellon University

Yang Gao
Tsinghua University

Agrim Gupta
Stanford University

David Held
Carnegie Mellon University

Chuer Pan
Stanford University

Daniel Seita
University of Southern California

Sanjana Srivastava
Stanford University