Decision & Control: Toward Fast and Accurate Robot Simulation with Multi-Contacts and Deformability

Importance of fast and physically-accurate robot simulation has never been more, as it is not only recognized as the key prerequisite for the recently-burgeoning learning-based methods in robotics but also the key building block for such applications as VR/AR, digital twins, etc.  In this talk, I will introduce some recent results in my lab on this topic: 1) PMI (passive mid-point integrator), which, by enforcing passivity in discrete-time domain, can stably and robustly simulate very light/stiff components in real-time while also allowing for variable update, information delay and modular construction of complex environments; 2) data-driven contact-intensive tight-tolerance simulation framework, which, by utilizing a network for contact clustering and mixed nodal/configuration-space contact models, can simulate such challenging tasks as bolt-fastening and peg-insertion tasks accurately and in real-time; and 3) COND (contact nodalization and diagonalization), which, by nodalizing all the contacts along with a certain surrogate dynamics, can decouple all the contacts (and their axes) with no matrix-inversion necessary, thereby, significantly enhancing speed and accuracy of simulation, particularly for large-dimensional dynamics with deformability.

Dongjun Lee (B.S. KAIST, 1995, M.S. KAIST, 1997, Ph.D. Minnesota 2004) is a Professor with the Department of Mechanical Engineering at the Seoul National University (SNU).  Prior to that, he was a Visiting Faculty at LAAS-CNRS, Toulouse, France, an Assistant Professor at the University of Tennessee – Knoxville from 2006 to 2011, a Postdoctoral Fellow with the Coordinated Science Laboratory at the UIUC from 2004 to 2006, and an Engine Development Engineer at Kia Motors, Korea, from 1997 to 1999.  His main research interests include mechanics and control of robotic and mechatronic systems with particular emphasis on aerial robotics and manipulation, teleoperation/haptics/VR, robot simulation, multirobot systems, and industrial control applications.  Dr. Lee is the recipient of US NSF CAREER Award 2009, the Best Paper Award from IAS 2012, the Best Video Award from URAI 2013 and the Best Paper Award from KRoC 2020.  He has been involved with various editorial services, including an Editor of ICRA (International Conference on Robotics & Automation), an Associate Editor of IEEE Trans. on Robotics, an Associate Editor of IEEE Trans. on Haptics and the Editor-in-Chief of IEEE UR (Ubiquitous Robots).

Web: https://www.inrol.snu.ac.kr/

Youtube: https://www.youtube.com/channel/UCk8xozp2_B-6qkN9XKzH_RA/videos