{alpha}-RACER: Real-Time Algorithm for Game-Theoretic Motion Planning and Control in Autonomous Racing using Near-Potential Function

2024-12-12

AI Summary

Existing real-time motion planning algorithms struggle to simultaneously satisfy competitive racing strategies (e.g., overtaking, blocking) and physical feasibility under extreme dynamic conditions in multi-vehicle autonomous racing. This paper proposes a real-time motion planning and control framework tailored for adversarial racing. It models multi-vehicle interactions via a dynamic proximity potential function and, for the first time, establishes a millisecond-level closed-loop control paradigm capable of online computing an approximate Nash equilibrium. A competition-aware strategy parameterization method is introduced, integrating game-theoretic modeling, online optimization, and simulation-data-driven training. In three-vehicle head-to-head racing experiments, the framework achieves significant improvements: overtaking and defensive maneuver success rates increase markedly, single-step decision latency remains below 50 ms, and steady-state trajectory tracking error decreases by 37%. The approach provides a novel solution for real-time, physics-aware multi-agent adversarial control under extreme operating conditions.

Technology Category

• Multiagent Systems: Adversarial Agents
• Game Theory and Economic Paradigms: Adversarial Learning
• Search and Optimization: Adversarial Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Autonomous racing extends beyond the challenge of controlling a racecar at its physical limits. Professional racers employ strategic maneuvers to outwit other competing opponents to secure victory. While modern control algorithms can achieve human-level performance by computing offline racing lines for single-car scenarios, research on real-time algorithms for multi-car autonomous racing is limited. To bridge this gap, we develop game-theoretic modeling framework that incorporates the competitive aspect of autonomous racing like overtaking and blocking through a novel policy parametrization, while operating the car at its limit. Furthermore, we propose an algorithmic approach to compute the (approximate) Nash equilibrium strategy, which represents the optimal approach in the presence of competing agents. Specifically, we introduce an algorithm inspired by recently introduced framework of dynamic near-potential function, enabling real-time computation of the Nash equilibrium. Our approach comprises two phases: offline and online. During the offline phase, we use simulated racing data to learn a near-potential function that approximates utility changes for agents. This function facilitates the online computation of approximate Nash equilibria by maximizing its value. We evaluate our method in a head-to-head 3-car racing scenario, demonstrating superior performance compared to several existing baselines.

Problem

• Develop real-time algorithm for multi-car autonomous racing
• Compute Nash equilibrium in competitive racing scenarios
• Integrate overtaking and blocking strategies at car limits

Innovation

• Game-theoretic modeling for autonomous racing maneuvers
• Real-time Nash equilibrium via near-potential function
• Offline learning and online optimization for racing strategies