Multi-Agent Adversarial Inverse Reinforcement Learning

Abstract

Reinforcement learning agents are prone to undesired behaviors due to reward mis-specification. Finding a set of reward functions to properly guide agent behaviors is particularly challenging in multi-agent scenarios. Inverse reinforcement learning provides a framework to automatically acquire suitable reward functions from expert demonstrations. Its extension to multi-agent settings, however, is difficult due to the more complex notions of rational behaviors. In this paper, we propose MA-AIRL, a new framework for multi-agent inverse reinforcement learning, which is effective and scalable for Markov games with high-dimensional state-action space and unknown dynamics. We derive our algorithm based on a new solution concept and maximum pseudolikelihood estimation within an adversarial reward learning framework. In the experiments, we demonstrate that MA-AIRL can recover reward functions that are highly correlated with ground truth ones, and significantly outperforms prior methods in terms of policy imitation.

arXiv commentICML 2019

Cited by in corpus (16)

• Modeling Human Driving Behavior through Generative Adversarial Imitation Learning
• MADRaS : Multi Agent Driving Simulator
• On Multi-Agent Learning in Team Sports Games
• Algorithms in Multi-Agent Systems: A Holistic Perspective from Reinforcement Learning and Game Theory
• Robust Reinforcement Learning via Adversarial training with Langevin Dynamics
• Grounded Relational Inference: Domain Knowledge Driven Explainable Autonomous Driving
• Scalable Multi-Agent Inverse Reinforcement Learning via Actor-Attention-Critic
• Semi-Supervised Imitation Learning of Team Policies from Suboptimal Demonstrations
• Multi-Agent Interactions Modeling with Correlated Policies
• Off-Policy Exploitability-Evaluation in Two-Player Zero-Sum Markov Games
• Bounded Risk-Sensitive Markov Games: Forward Policy Design and Inverse Reward Learning with Iterative Reasoning and Cumulative Prospect Theory
• Learning Altruistic Behaviours in Reinforcement Learning without External Rewards
• Decentralized Multi-Agents by Imitation of a Centralized Controller
• Multi-Agent Imitation Learning with Copulas
• Bridging the Gap Between $f$-GANs and Wasserstein GANs
• FireCommander: An Interactive, Probabilistic Multi-agent Environment for Heterogeneous Robot Teams