Minimum Coverage Sets for Training Robust Ad Hoc Teamwork Agents

TL;DR

This paper introduces L-BRDiv, an algorithm that trains robust ad hoc teamwork agents by generating teammate policies that emulate the minimum coverage set, improving robustness across various cooperative problems.

Contribution

The paper proposes the L-BRDiv algorithm to generate teammate policies that better approximate the minimum coverage set, enhancing robustness in ad hoc teamwork agents.

Findings

L-BRDiv produces more robust agents than state-of-the-art methods.

L-BRDiv does not require extensive hyperparameter tuning.

It outperforms baselines by discovering diverse MCS members.

Abstract

Robustly cooperating with unseen agents and human partners presents significant challenges due to the diverse cooperative conventions these partners may adopt. Existing Ad Hoc Teamwork (AHT) methods address this challenge by training an agent with a population of diverse teammate policies obtained through maximizing specific diversity metrics. However, prior heuristic-based diversity metrics do not always maximize the agent's robustness in all cooperative problems. In this work, we first propose that maximizing an AHT agent's robustness requires it to emulate policies in the minimum coverage set (MCS), the set of best-response policies to any partner policies in the environment. We then introduce the L-BRDiv algorithm that generates a set of teammate policies that, when used for AHT training, encourage agents to emulate policies from the MCS. L-BRDiv works by solving a constrained optimization problem to jointly train teammate policies for AHT training and approximating AHT agent policies that are members of the MCS. We empirically demonstrate that L-BRDiv produces more robust AHT agents than state-of-the-art methods in a broader range of two-player cooperative problems without the need for extensive hyperparameter tuning for its objectives. Our study shows that L-BRDiv outperforms the baseline methods by prioritizing discovering distinct members of the MCS instead of repeatedly finding redundant policies.