GTP-Force: Game-Theoretic Trajectory Prediction through Distributed Reinforcement Learning

This paper introduces Game-theoretic Trajectory Prediction through distributed reinForcement learning (GTP-Force), a system that tackles the challenge of predicting joint pedestrian trajectories in multi-agent scenarios. GTP-Force utilizes decentralized reinforcement learning agents to personalize neural networks for each competing player based on their non-cooperative preferences and social interactions with others. By identifying the Nash Equilibria, GTP-Force accurately predicts joint trajectories while minimizing overall system loss in non-cooperative environments. The system outperforms existing state-of-the-art trajectory predictors, achieving an average displacement error of 0.19m on the ETH+UCY dataset and 80% accuracy on the Orange dataset, which is -0.01m and 5% better than the best-performing baseline, respectively. Additionally, GTP-Force considerably reduces the model size of social mobility predictors compared to approaches with classical game theory.

I. Introduction

Forecasting mobile users’ motion patterns, whether pedestrians or vehicles, has become increasingly important in urban planning and intelligent mobility systems. It can enable various technologies, such as intelligent transportation services, safety and emergency applications, rescue operations, autonomous vehicles, and road traffic engineering. Similarly, mobility prediction is a pivotal aspect of enabling various wireless network applications, including adaptive and anticipatory network management, resource allocation, handover management, and proactive service migration [5].