Multi-Agent Manipulation via Locomotion using Hierarchical Sim2Real

TL;DR

This paper presents a hierarchical sim2real reinforcement learning approach enabling multiple mobile agents to coordinate for manipulation tasks, successfully transferring learned behaviors from simulation to real-world without additional training.

Contribution

The work introduces a hierarchical sim2real framework that combines low-level goal-reaching skills with high-level RL controllers for multi-agent manipulation, facilitating zero-shot transfer to real environments.

Findings

Successful real-world multi-agent manipulation demonstrations

Effective generalization through domain randomization

Modular hierarchy improves learning and transfer of complex behaviors

Abstract

Manipulation and locomotion are closely related problems that are often studied in isolation. In this work, we study the problem of coordinating multiple mobile agents to exhibit manipulation behaviors using a reinforcement learning (RL) approach. Our method hinges on the use of hierarchical sim2real -- a simulated environment is used to learn low-level goal-reaching skills, which are then used as the action space for a high-level RL controller, also trained in simulation. The full hierarchical policy is then transferred to the real world in a zero-shot fashion. The application of domain randomization during training enables the learned behaviors to generalize to real-world settings, while the use of hierarchy provides a modular paradigm for learning and transferring increasingly complex behaviors. We evaluate our method on a number of real-world tasks, including coordinated object manipulation in a multi-agent setting. See videos at https://sites.google.com/view/manipulation-via-locomotion