MPOGames: Efficient Multimodal Partially Observable Dynamic Games

TL;DR

MPOGames introduces an efficient method for solving multimodal, partially observable dynamic games by capturing multiple local Nash equilibria, enabling better handling of uncertainty in multi-agent interactions in real-time scenarios.

Contribution

The paper presents MPOGames, a novel approach reformulating MaxEnt dynamic games as POMDPs to efficiently address multiple local Nash equilibria and uncertainty in multi-agent planning.

Findings

Effective in modeling multiple local Nash equilibria.

Demonstrates real-time performance on hardware platform.

Improves planning accuracy in multi-agent scenarios.

Abstract

Game theoretic methods have become popular for planning and prediction in situations involving rich multi-agent interactions. However, these methods often assume the existence of a single local Nash equilibria and are hence unable to handle uncertainty in the intentions of different agents. While maximum entropy (MaxEnt) dynamic games try to address this issue, practical approaches solve for MaxEnt Nash equilibria using linear-quadratic approximations which are restricted to unimodal responses and unsuitable for scenarios with multiple local Nash equilibria. By reformulating the problem as a POMDP, we propose MPOGames, a method for efficiently solving MaxEnt dynamic games that captures the interactions between local Nash equilibria. We show the importance of uncertainty-aware game theoretic methods via a two-agent merge case study. Finally, we prove the real-time capabilities of our approach with hardware experiments on a 1/10th scale car platform.