Overcooked Generalisation Challenge

New AI method enables robust ad-hoc teamwork

Researchers have developed a new multi-agent reinforcement learning method called Unsupervised Partner Design (UPD). This technique generates training partners dynamically during the learning process, adapting them base…