Cooperative Open-ended Learning Framework for Zero-shot Coordination
This addresses cooperative incompatibility in AI systems for scenarios requiring coordination with unseen partners, representing a novel method for a known bottleneck.
The paper tackles the challenge of zero-shot coordination in cooperative AI by proposing the COLE framework, which uses open-ended objectives and graph theory to assess strategies, and demonstrates that it outperforms state-of-the-art methods in the Overcooked game environment.
Zero-shot coordination in cooperative artificial intelligence (AI) remains a significant challenge, which means effectively coordinating with a wide range of unseen partners. Previous algorithms have attempted to address this challenge by optimizing fixed objectives within a population to improve strategy or behaviour diversity. However, these approaches can result in a loss of learning and an inability to cooperate with certain strategies within the population, known as cooperative incompatibility. To address this issue, we propose the Cooperative Open-ended LEarning (COLE) framework, which constructs open-ended objectives in cooperative games with two players from the perspective of graph theory to assess and identify the cooperative ability of each strategy. We further specify the framework and propose a practical algorithm that leverages knowledge from game theory and graph theory. Furthermore, an analysis of the learning process of the algorithm shows that it can efficiently overcome cooperative incompatibility. The experimental results in the Overcooked game environment demonstrate that our method outperforms current state-of-the-art methods when coordinating with different-level partners. Our demo is available at https://sites.google.com/view/cole-2023.