Abstract
Autonomous racing has gained significant attention as a platform forhigh-speed decision-making and motion control. While existing methods primarilyfocus on trajectory planning and overtaking strategies, the role ofsportsmanship in ensuring fair competition remains largely unexplored. In humanracing, rules such as the one-motion rule and the enough-space rule preventdangerous and unsportsmanlike behavior. However, autonomous racing systemsoften lack mechanisms to enforce these principles, potentially leading tounsafe maneuvers. This paper introduces a bi-level game-theoretic framework tointegrate sportsmanship (SPS) into versus racing. At the high level, we modelracing intentions using a Stackelberg game, where Monte Carlo Tree Search(MCTS) is employed to derive optimal strategies. At the low level, vehicleinteractions are formulated as a Generalized Nash Equilibrium Problem (GNEP),ensuring that all agents follow sportsmanship constraints while optimizingtheir trajectories. Simulation results demonstrate the effectiveness of theproposed approach in enforcing sportsmanship rules while maintainingcompetitive performance. We analyze different scenarios where attackers anddefenders adhere to or disregard sportsmanship rules and show how knowledge ofthese constraints influences strategic decision-making. This work highlightsthe importance of balancing competition and fairness in autonomous racing andprovides a foundation for developing ethical and safe AI-driven racing systems.