Exciting Contact Modes in Differentiable Simulations for Robot Learning

2024/11/17 by Sathyanarayan, Hrishikesh, Abraham, Ian · 1 citation
#FOS: Computer and information sciences #Information Theory (cs.IT) #Robotics (cs.RO)

paper · doi:10.48550/arxiv.2411.10935

Abstract

In this paper, we explore an approach to actively plan and excite contact modes in differentiable simulators as a means to tighten the sim-to-real gap. We propose an optimal experimental design approach derived from information-theoretic methods to identify and search for information-rich contact modes through the use of contact-implicit optimization. We demonstrate our approach on a robot parameter estimation problem with unknown inertial and kinematic parameters which actively seeks contacts with a nearby surface. We show that our approach improves the identification of unknown parameter estimates over experimental runs by an estimate error reduction of at least ∼ 84% when compared to a random sampling baseline, with significantly higher information gains.

Cited by

• Sampling-Based System Identification with Active Exploration for Legged Robot Sim2Real Learning

Related

• Behavior Synthesis via Contact-Aware Fisher Information Maximization
• Efficient Differentiable Contact Model with Long-range Influence
• Contact Optimization with Learning from Demonstration: Application in Long-term Non-prehensile Planar Manipulation
• Simultaneous Learning of Contact and Continuous Dynamics
• Information-theoretic Model Identification and Policy Search using Physics Engines with Application to Robotic Manipulation