Unsupervised Partner Design Enables Robust Ad-hoc Teamwork

TL;DR

This paper presents Unsupervised Partner Design (UPD), a novel reinforcement learning framework that adaptively generates diverse training partners for robust ad-hoc teamwork without pretraining or manual tuning.

Contribution

UPD introduces a population-free, stochastic partner generation method combined with unsupervised environment design for fully unsupervised curricula in cooperative multi-agent settings.

Findings

UPD outperforms baselines in Overcooked-AI and generalization challenges.

UPD achieves higher user-rated adaptability and human-likeness.

The method enables fully unsupervised curriculum learning over level and partner distributions.

Abstract

We introduce Unsupervised Partner Design (UPD) - a population-free, multi-agent reinforcement learning framework for robust ad-hoc teamwork that adaptively generates training partners without requiring pretrained partners or manual parameter tuning. UPD constructs diverse partners by stochastically mixing an ego agent's policy with biased random behaviours and scores them using a variance-based learnability metric that prioritises partners near the ego agent's current learning frontier. We show that UPD can be integrated with unsupervised environment design, resulting in the first method enabling fully unsupervised curricula over both level and partner distributions in a cooperative setting. Through extensive evaluations on Overcooked-AI and the Overcooked Generalisation Challenge, we demonstrate that this dynamic partner curriculum is highly effective: UPD consistently outperforms both population-based and population-free baselines as well as ablations. In a user study, we further show that UPD achieves higher returns than all baselines and was perceived as significantly more adaptive, more human-like, a better collaborator, and less frustrating.