Generative adversarial imitation learning for robot swarms: Learning from human demonstrations and trained policies

TL;DR

This paper introduces a generative adversarial imitation learning framework for swarm robotics, enabling robots to learn collective behaviors from human demonstrations and trained policies, with successful real-world deployment.

Contribution

It presents a novel imitation learning approach for swarm robotics that learns from human and policy-derived demonstrations, validated through real-robot experiments.

Findings

Behaviors learned are qualitatively meaningful and similar to demonstrations.

Policies perform comparably in simulation and real-world experiments.

The framework effectively transfers learned behaviors to physical robot swarms.

Abstract

In imitation learning, robots are supposed to learn from demonstrations of the desired behavior. Most of the work in imitation learning for swarm robotics provides the demonstrations as rollouts of an existing policy. In this work, we provide a framework based on generative adversarial imitation learning that aims to learn collective behaviors from human demonstrations. Our framework is evaluated across six different missions, learning both from manual demonstrations and demonstrations derived from a PPO-trained policy. Results show that the imitation learning process is able to learn qualitatively meaningful behaviors that perform similarly well as the provided demonstrations. Additionally, we deploy the learned policies on a swarm of TurtleBot 4 robots in real-robot experiments. The exhibited behaviors preserved their visually recognizable character and their performance is comparable to the one achieved in simulation.