UBSoft: A Simulation Platform for Robotic Skill Learning in Unbounded Soft Environments

TL;DR

UBSoft, a new simulation platform designed to support unbounded soft environments for robot skill acquisition, is introduced, which markedly reduces the demand for extensive storage space and computation costs required for large-scale scenarios involving soft materials.

Abstract

It is desired to equip robots with the capability of interacting with various soft materials as they are ubiquitous in the real world. While physics simulations are one of the predominant methods for data collection and robot training, simulating soft materials presents considerable challenges. Specifically, it is significantly more costly than simulating rigid objects in terms of simulation speed and storage requirements. These limitations typically restrict the scope of studies on soft materials to small and bounded areas, thereby hindering the learning of skills in broader spaces. To address this issue, we introduce UBSoft, a new simulation platform designed to support unbounded soft environments for robot skill acquisition. Our platform utilizes spatially adaptive resolution scales, where simulation resolution dynamically adjusts based on proximity to active robotic agents. Our framework markedly reduces the demand for extensive storage space and computation costs required for large-scale scenarios involving soft materials. We also establish a set of benchmark tasks in our platform, including both locomotion and manipulation tasks, and conduct experiments to evaluate the efficacy of various reinforcement learning algorithms and trajectory optimization techniques, both gradient-based and sampling-based. Preliminary results indicate that sampling-based trajectory optimization generally achieves better results for obtaining one trajectory to solve the task. Additionally, we conduct experiments in real-world environments to demonstrate that advancements made in our UBSoft simulator could translate to improved robot interactions with large-scale soft material. More videos can be found at https://vis-www.cs.umass.edu/ubsoft/.

Figures (9)

• Figure 1: The title of the paper is written on a large scene covered with soft material by a robotic manipulator.
• Figure 2: MPM Simulation with Spatially Adaptive Scheme. Hierarchical grids are centered around the robot agent and become sparser further out. As the robot moves, these grids move in sync, and the particles are resampled to match the grid cell sizes by splitting large particles in small grids and merging small particles in large grids. Subsequent phases of P2G and G2P adhere to standard MPM protocols but are adapted to accommodate hierarchical grids and particles.
• Figure 3: 4 manipulation tasks and 4 locomotion tasks proposed in UBSoft.
• Figure 4: With larger scenes, the spatially adaptive scheme significantly reduces the storage space required and maintains a higher simulation speed. The dotted lines are extrapolated results where transitional MPM fails to simulate.
• Figure 5: Reward curves for all methods including PPO, SAC, CMA-ES, and Differentiable Physics.