Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation
This work addresses the problem of sample complexity in RL for robotics with deformable materials, which is incremental as it builds on existing differentiable simulation and RL methods.
The paper tackles the challenge of scaling reinforcement learning (RL) to tasks involving soft bodies and deformables, which are slower to simulate than rigid bodies, by introducing a novel RL algorithm and a differentiable simulation platform, showing that their approach outperforms baselines on manipulation and locomotion tasks.
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. Additional details at https://rewarped.github.io/.