The 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024), one of the premier global conferences on robotics research, took place from October 14 to October 18, 2024, at the Abu Dhabi National Exhibition Centre (ADNEC) in Abu Dhabi, UAE.
A paper titled “A Convex Formulation of Frictional Contact for the Material Point Method and Rigid Bodies” was honored as one of the five Finalists for the IROS 2024 RoboCup Best Paper Award. Authored by Zeshun Zong, a fifth-year Ph.D. student in the UCLA Department of Mathematics; Associate Professor Chenfanfu Jiang; and Dr. Xuchen Han, a research scientist at Toyota Research Institute and UCLA Math Department Ph.D. alumnus (Class of 2020), the paper introduces a groundbreaking approach to frictional contact modeling in robotic simulation. The method significantly advances the stability and robustness of simulations involving complex contact scenarios, which is crucial for improving robotic manipulation tasks.
“This paper presents a novel convex formulation that seamlessly integrates the Material Point Method (MPM) with rigid body dynamics under frictional contact. This allows for robust simulations, essential for testing and refining robotic control policies in challenging environments,” said Professor Jiang.
“Our work is centered on ensuring the mathematical stability and computational efficiency required for accurately simulating these complex interactions. By leveraging a constrained optimization framework, we were able to maintain a high degree of accuracy while making the solver scalable for practical use cases,” added Zeshun Zong, the first author and a fifth-year Ph.D. student at Jiang’s lab at UCLA.
The complete paper can be accessed on arXiv: https://arxiv.org/abs/2403.13783, and a supplemental video is available on YouTube: https://www.youtube.com/watch?v=5jrQtF5D0DA.
This collaboration is a result of a joint multi-year sponsored project from Toyota Research Institute (TRI) to Professor Jiang’s research group. The joint project, titled Robust Accurate and Interactive Solutions for Contact-Rich Multi-Physics Robotic Simulations, is led by Professor Jiang as the Principal Investigator (PI) at UCLA, with TRI co-PIs including research scientists Xuchen Han, Joe Masterjohn, and Alejandro Castro. The project focuses on advancing numerical methods and computational continuum mechanics, particularly for robotics and AI applications.
“This collaboration is crucial for advancing robotic simulation capabilities at TRI. It allows us to bring theoretical advancements into real-world robotics, pushing the boundaries of what we can achieve in simulation technology,” said Dr. Xuchen Han.
Professor Jiang directs the UCLA Artificial Intelligence and Visual Computing (AIVC) Laboratory. The lab’s research focuses on the synergy between visual computing and embodied artificial intelligence. The lab is dedicated to advancing technologies in the creation, representation, understanding, and optimization of digital 3D content and intelligent robots. The TRI collaboration involves multiple research thrusts aimed at improving simulation accuracy, efficiency, and real-world applicability. The collaboration aims to bridge the gap between advanced simulation methodologies and practical robotic applications by integrating these innovations into open-source toolkits, thereby fostering community involvement and setting new standards in robotics simulation technology.
