ArrayBot: Reinforcement Learning for Generalizable Distributed Manipulation through Touch

TL;DR

ArrayBot employs reinforcement learning on a tactile sensor array to develop generalizable distributed manipulation policies that transfer seamlessly from simulation to real-world tasks without domain randomization.

Contribution

This work introduces a novel tactile-array-based manipulation system and a reshaped RL action space enabling generalizable and transferable distributed manipulation policies.

Findings

Policies generalize to unseen object shapes in simulation

Policies transfer to physical robots without domain randomization

ArrayBot successfully performs diverse real-world manipulation tasks

Abstract

We present ArrayBot, a distributed manipulation system consisting of a $16 \times 16$ array of vertically sliding pillars integrated with tactile sensors, which can simultaneously support, perceive, and manipulate the tabletop objects. Towards generalizable distributed manipulation, we leverage reinforcement learning (RL) algorithms for the automatic discovery of control policies. In the face of the massively redundant actions, we propose to reshape the action space by considering the spatially local action patch and the low-frequency actions in the frequency domain. With this reshaped action space, we train RL agents that can relocate diverse objects through tactile observations only. Surprisingly, we find that the discovered policy can not only generalize to unseen object shapes in the simulator but also transfer to the physical robot without any domain randomization. Leveraging the deployed policy, we present abundant real-world manipulation tasks, illustrating the vast potential of RL on ArrayBot for distributed manipulation.