Cooperative Opponent Intention Recognition Based on Trajectory Analysis by Distributed Fractional Bayesian Inference
Peng Yi, Yaqun Yang, Zixiang Lin · 2024
Opponent intention recognition is a crucial component for multi-agent decision-making. This paper proposes a novel distributed fractional Bayesian inference algorithm for opponent intention recognition in the Multiagent Pursuit-Evasion (MPE) game for Unmanned Aerial Vehicle (UAV) confrontation. The algorithm infers the opponent's intention between pursuit and evasion by analyzing the trajectories in dynamic interaction under a game-theoretical model. This process involves comparing the prediction trajectory with the observed noisy trajectory to update the intention beliefs using fractional Bayesian inference. Additionally, it obtains a consensual intention through distributed linear average. Our method can effectively distinguish the intention of opponents in MPE games. Experimental validation demonstrates that the proposed algorithm performance.