A Behavior-based Approach for Multi-agent Q-learning for Autonomous Exploration

TL;DR

This paper presents a multi-agent reinforcement learning approach using behavior-based architecture to enhance autonomous exploration capabilities of mobile robots in hazardous and unknown environments.

Contribution

It introduces a novel multi-agent reinforcement learning method with a behavior-based architecture for autonomous robot exploration, validated in real indoor and outdoor settings.

Findings

Effective exploration in hazardous environments

Successful implementation of multi-agent RL in real-world tests

Improved learning efficiency through behavior prioritization

Abstract

The use of mobile robots is being popular over the world mainly for autonomous explorations in hazardous/ toxic or unknown environments. This exploration will be more effective and efficient if the explorations in unknown environment can be aided with the learning from past experiences. Currently reinforcement learning is getting more acceptances for implementing learning in robots from the system-environment interactions. This learning can be implemented using the concept of both single-agent and multiagent. This paper describes such a multiagent approach for implementing a type of reinforcement learning using a priority based behaviour-based architecture. This proposed methodology has been successfully tested in both indoor and outdoor environments.