The Science of Bumping Into Things

Towards Robots That Aren’t Afraid of Contact

Workshop at Robotics: Science and Systems 2022

July 1st, 2022

Description

Handling contact is a fundamental challenge in robotics. We want robots to interact with the world to do something useful, which often requires physical interaction with the environment. Whether it is a warehouse arm picking an object from a box, a search and rescue robot walking over broken terrain, or a home assistance robot clearing the dishes, robots need to be able to enable, plan for, control through, sense, withstand, and learn about contact events. However, contact necessarily involves a sudden change in dynamics – the effects of motion while touching and almost touching can differ dramatically. Because of this, many robots seek to avoid or minimize the effects of contact, for example by slowing down before touching an object. Furthermore, robots go through great lengths to avoid unexpected contacts, by operating in controlled environments or taking excessively conservative paths around obstacles.

This workshop seeks to push the field toward a future where robots aren’t afraid of contact. Where accidentally bumping into something isn’t catastrophic, and where the robot doesn’t treat everything in the environment as if it were made of egg shells. To get there, we need to reason about contact while considering all aspects of the system design: the mechanism, sensing, control, planning, and learning must all account for the challenges of real-world contact. This is also an issue that cuts across different application domains such as manipulation or locomotion. Thus this workshop is designed to bring together researchers from across the robotics field to explore new approaches to tackling this fundamental challenge.

To get the conversation started, here is a list of topics and questions for participants to think about (but other topics within the spirit of the above description will also be considered):

  • Reactive or online planning and control to initiate and leverage contacts.
  • Robustness to expected or unexpected contact events.
  • Is a model-based structure necessary, or can unstructured learning methods succeed? What types of problems might require or benefit from structure?
  • Making sense of touch: leveraging spatially rich tactile sensing.
  • The role of design and novel shapes to control interactions.
  • Interacting with soft and deformable surfaces, and leveraging softness for robustness.
  • How does the manner (fast/slow, accidental/intentional/adversarial) or physics (soft/hard, point/surface, slippery/sticky) affect which solution(s) might be appropriate?

Video Recording of the Workshop

Invited Speakers

We are excited to have an incredible group of invited speakers from a range of research backgrounds who are all excited to share their thoughts on this theme:

Call for Contributions

We would like to invite contributed abstracts from the community that fit the themes of this workshop. Accepted abstracts will be presented as either posters or spotlight talks. Submissions may be on recently published work, upcoming papers, position papers, or work in progress. All accepted abstracts will be posted on this page. Submissions should follow the RSS paper format and should be one to two pages in length (including citations). Submissions should include a brief abstract at the start summarizing the work.

Submission deadline has past

Schedule

Location: 451 CSB, Posters: CS Lounge

Introduction & Welcome 9:00-9:15  
Maria Bauza 9:15-9:35 Video
Ken Goldberg 9:35-9:55 Video
Carolyn Matl 9:55-10:15  
Huanbo Sun 10:15-10:30 Video
Posters and coffee break 10:30-11:30  
Kevin Lynch 11:30-11:50 Video
Panel Discussion 1 11:50-12:20 Video
Lunch 12:20-1:50  
Taylor Howell and Simon Le Cleac’h 1:50-2:05 Video
Yuki Shirai 2:05-2:20 Video
Francesca Negrello 2:20-2:40  
Yan Gu 2:40-3:00 Video
Coffee break 3:00-3:30  
Alessandro Saccon 3:30-3:50 Video
Zachary Manchester 3:50-4:10 Video
Pulkit Agrawal 4:10-4:30 Video
Panel Discussion 2 4:30-5:00 Video

Contributed Papers

The accepted abstracts for this workshop are:
Fast Contact-Implicit Model-Predictive Control, by Howell, Taylor; Le Cleac’h, Simon; Schwager, Mac; and Manchester, Zachary
Insight: A Vision-based Haptic Sensor That Provides Rich Force Information about Contacts, by Sun, Huanbo; Kuchenbecker, Katherine J; and Martius, Georg
Safe Supervisory Control of Soft Robot Actuators, by Sabelhaus, Andrew P; Patterson, Zach; Wertz, Anthony; and Majidi, Carmel
Framework and Software for Real-Time Multi-Contact Model Predictive Control, by Aydinoglu, Alp; and Posa, Michael
Robust Pivoting Manipulation Using Bilevel Contact-Implicit Optimization, by Shirai, Yuki; Jha, Devesh K; Raghunathan, Arvind U; and Romeres, Diego
Efficient Object Manipulation Planning with Monte Carlo Tree Search, by Zhu, Huaijiang; and Righetti, Ludovic
Using Contacts During Robot Manipulation to Map and Reconstruct a Scene, by Bimbo, Joao; Morgan, Andrew S; and Dollar, Aaron M.
Python-based Open Source Package for Optimization of Contact-rich Systems, by Raghunathan, Arvind U; Jha, Devesh K; and Romeres, Diego
Developing Data-driven Methods for Mixed-integer Nonlinear Programs, by Lin, Xuan; and Hong, Dennis W
Dynamic Inference on Graphs using Structured Transition Models, by Saxena, Saumya; and Kroemer, Oliver
Torque-Limited Manipulation Planning with Contact, by Natarajan, Ramkumar; Likhachev, Maxim; and Choset, Howie
Emergent Mechanism Design via Robot Swarms, by Taylor, Annalisa; Berrueta, Thomas A; and Murphey, Todd
High-Speed Scooping Manipulation Using Controlled Compliance, by Mak, Ka Hei; and Seo, Jungwon
Hybrid Event Shaping to Stabilize Periodic Hybrid Orbits, by Zhu, James; Kong, Nathan J; Council, George; and Johnson, Aaron M
Solving the “Last Centimeter” Problem with Autonomous Grasping Reflexes, by SaLoutos, Andrew; Guo, Menglong; Stanger-Jones, Elijah; Kim, Hongmin; and Kim, Sangbae
Automatic Tool Design for Robotic Caging Using Flexible Wires, by Ketchum, Jake B; Meyer, Joel; and Murphey, Todd
Designing Whisker Sensors for Noisy Environments, by Kent, Teresa A; Kim, Suhan; Babaei, Mahnoush; Emnett, Hannah; Hartmann, Mitra J. Z.; and Bergbreiter, Sarah

Organizers