Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation

Recent advances in GPU-based parallel simulation have enabled practitioners to collect large amounts of data and train complex control policies using deep reinforcement learning (RL), on commodity GPUs. However, such successes for RL in robotics have been limited to tasks sufficiently simulated by fast rigid-body dynamics. Simulation techniques for soft bodies are comparatively several orders of magnitude slower, thereby limiting the use of RL due to sample complexity requirements. To address this challenge, this paper presents both a novel RL algorithm and a simulation platform to enable scaling RL on tasks involving rigid bodies and deformables. We introduce Soft Analytic Policy Optimization (SAPO), a maximum entropy first-order model-based actor-critic RL algorithm, which uses first-order analytic gradients from differentiable simulation to train a stochastic actor to maximize expected return and entropy. Alongside our approach, we develop Rewarped, a parallel differentiable multiphysics simulation platform that supports simulating various materials beyond rigid bodies. We re-implement challenging manipulation and locomotion tasks in Rewarped, and show that SAPO outperforms baselines over a range of tasks that involve interaction between rigid bodies, articulations, and deformables. iii I am grateful to my advisor, Jean Oh, for her guidance and unwavering support. The path taken to complete this work has been filled with many twists and turns. I am incredibly excited with where we ended up, and it would not have lead to this with anyone else. To my committee members-Dave Held, Jeff Ichnowski, Uksang Yoo-I am thankful for your insightful feedback and discussions that helped shape my work. Additionally, it has been a privilege to work with Vernon Luk, and I hope my mentorship has made a positive influence on your future aspirations. Furthermore, I would like to extend a heartfelt thank you to peers and friends who have made my time in Pittsburgh to now unforgettable-