Diverse agents in Hanabi
Hanabi is a cooperative card game made by Antoine Bauza, where players attempt to play cards in the correct order according to their values and color.
The catch: players don’t see their own hand and can only communicate through a limited number of hints by telling other players about their cards. For this reason, Hanabi has been proposed as an AI benchmark for agents that reason about other player’s beliefs and intentions.
The aim of this project is to leverage algorithms that can generate a large number of behaviorally diverse agents, such as MAP-Elites, to search for robust and adaptive strategies to play well with as many different agents as possible.
Publications
- Diverse agents for ad-hoc cooperation in Hanabi (CoG 2019)
- Evolving agents for the Hanabi 2018 CIG competition (CIG 2018)
- Behavioral Evaluation Rainbow Rulebased Hanabi (AIIDE 2020)