Introduction to Multi-Agent Reinforcement Learning with Petting Zoo

Speaker : Claire Bizon Monroc
Inria
Date: 16/11/2022
Time: 11:30 am - 12:30 pm
Location: Telecom Paris

Abstract

PettingZoo (https://pettingzoo.farama.org/) is an open source Python source library supported by Farama Foundation to accelerate research in multi-agent reinforcement learning (RL). Inheriting the standard features of the popular RL interface Gym, it extends its API to handle multiple agents by following an Agent Environment Cycle (“AEC”) games mode. In this workshop we will:

  • Discover the PettingZoo library and explore supported multi-agent environments,
  • Learn RL policies for a multi-agent benchmark task in a PettingZoo environment,
  • Create a custom environment in PettingZoo.