ReCollab: Retrieval-Augmented LLMs for Cooperative Ad-hoc Teammate Modeling
This addresses the challenge of coordinating with unseen teammates in cooperative AI settings, representing an incremental improvement over existing language-based methods.
The paper tackled the problem of ad-hoc teamwork by using large language models to infer teammate behavior from short traces, introducing ReCollab with retrieval-augmented generation to improve adaptation, achieving Pareto-optimal trade-offs in classification accuracy and episodic return in the Overcooked environment.
Ad-hoc teamwork (AHT) requires agents to infer the behavior of previously unseen teammates and adapt their policy accordingly. Conventional approaches often rely on fixed probabilistic models or classifiers, which can be brittle under partial observability and limited interaction. Large language models (LLMs) offer a flexible alternative: by mapping short behavioral traces into high-level hypotheses, they can serve as world models over teammate behavior. We introduce \Collab, a language-based framework that classifies partner types using a behavior rubric derived from trajectory features, and extend it to \ReCollab, which incorporates retrieval-augmented generation (RAG) to stabilize inference with exemplar trajectories. In the cooperative Overcooked environment, \Collab effectively distinguishes teammate types, while \ReCollab consistently improves adaptation across layouts, achieving Pareto-optimal trade-offs between classification accuracy and episodic return. These findings demonstrate the potential of LLMs as behavioral world models for AHT and highlight the importance of retrieval grounding in challenging coordination settings.