Enhancing Autonomous Racing Strategies: A Cognitive Hierarchy-Based Safe Motion Planning Approach

Autonomous vehicle motion planning in competitive scenarios, such as car racing, presents significant challenges due to the unpredictability of adversaries' behaviors. To generate intelligent autonomous behavior, vehicles must anticipate and react to the maneuvers of other vehicles. However, existing game-theoretic motion planning approaches often assume full knowledge of opponents' behavior patterns, overlooking the inherent uncertainty of such behaviors. This assumption can lead to suboptimal performance in critical maneuvers, such as blocking and overtaking. To address this issue, this paper proposes a motion planning method based on cognitive hierarchy theory. This method enhances the understanding of agent behavior patterns through a probabilistic model that updates inferences about each opponent's cognitive level from interaction feedback. Additionally, the proposed planner incorporates a new cost function and discrete-time control barrier functions to ensure safety during competition. The effectiveness of our planner is demonstrated through comparative simulations with the sensitivity-enhanced best response iteration (SE-IBR) algorithm. The results indicate that the proposed algorithm outperforms the SE-IBR in blocking and overtaking scenarios, highlighting its potential to improve autonomous strategies in competitive driving situations.