Emergent Tool Use From Multi-Agent Autocurricula

TL;DR

This paper demonstrates that multi-agent competition with simple objectives and reinforcement learning induces complex, emergent behaviors including tool use and coordination, with potential scalability and relevance to human skills.

Contribution

It reveals six distinct emergent strategy phases driven by self-supervised autocurricula in multi-agent hide-and-seek, highlighting complex tool use and coordination.

Findings

Agents develop multi-object shelters using moveable boxes

Agents discover obstacle overcoming with ramps

Multi-agent competition scales better with environment complexity

Abstract

Through multi-agent competition, the simple objective of hide-and-seek, and standard reinforcement learning algorithms at scale, we find that agents create a self-supervised autocurriculum inducing multiple distinct rounds of emergent strategy, many of which require sophisticated tool use and coordination. We find clear evidence of six emergent phases in agent strategy in our environment, each of which creates a new pressure for the opposing team to adapt; for instance, agents learn to build multi-object shelters using moveable boxes which in turn leads to agents discovering that they can overcome obstacles using ramps. We further provide evidence that multi-agent competition may scale better with increasing environment complexity and leads to behavior that centers around far more human-relevant skills than other self-supervised reinforcement learning methods such as intrinsic motivation. Finally, we propose transfer and fine-tuning as a way to quantitatively evaluate targeted capabilities, and we compare hide-and-seek agents to both intrinsic motivation and random initialization baselines in a suite of domain-specific intelligence tests.