Natural Emergence of Heterogeneous Strategies in Artificially Intelligent Competitive Teams

TL;DR

This paper introduces a multi-agent competitive environment where homogeneous agents naturally develop diverse strategies through reinforcement learning, enhancing team performance and flexibility.

Contribution

It demonstrates the emergence of heterogeneous behaviors among homogeneous agents in a competitive setting and proposes ensemble training for adaptable policies.

Findings

Heterogeneous strategies emerge naturally among homogeneous agents.

Graph Neural Networks with Reinforcement Learning facilitate complex strategy evolution.

Ensemble training produces versatile policies capable of replacing individual agents.

Abstract

Multi agent strategies in mixed cooperative-competitive environments can be hard to craft by hand because each agent needs to coordinate with its teammates while competing with its opponents. Learning based algorithms are appealing but many scenarios require heterogeneous agent behavior for the team's success and this increases the complexity of the learning algorithm. In this work, we develop a competitive multi agent environment called FortAttack in which two teams compete against each other. We corroborate that modeling agents with Graph Neural Networks and training them with Reinforcement Learning leads to the evolution of increasingly complex strategies for each team. We observe a natural emergence of heterogeneous behavior amongst homogeneous agents when such behavior can lead to the team's success. Such heterogeneous behavior from homogeneous agents is appealing because any agent can replace the role of another agent at test time. Finally, we propose ensemble training, in which we utilize the evolved opponent strategies to train a single policy for friendly agents.