SGTP: Sampling-based Game-Theoretic Planning for Real-Time Multi-Vehicle Autonomous Racing

2026-07-28

AI Summary

This work addresses the challenge of balancing strategic diversity and real-time computational efficiency in multi-agent autonomous racing under highly interactive scenarios. The authors propose a real-time planning framework that integrates game-theoretic reasoning with GPU-accelerated control sequence sampling. Candidate trajectories are ranked using game-aware costs, while track boundaries and dynamic collision-avoidance constraints are explicitly embedded to guarantee the safety and feasibility of strategy switching. By uniquely combining game-aware sampling with efficient feasibility filtering, the method achieves a win rate of 95.24% and a task completion rate of 99.35% in complex track simulations, with an average planning time of only 0.095 seconds. Furthermore, it scales successfully to large-scale scenarios involving ten vehicles, effectively reconciling diversity, safety, and real-time performance.

Abstract

Autonomous multi-vehicle racing requires real-time planning of diverse competitive behaviors in intense interactions. Existing planners often struggle to balance strategic diversity and computational efficiency. To address this challenge, we propose Sampling-based Game-Theoretic Planning (SGTP), a real-time framework that combines game-theoretic reasoning with GPU-accelerated sampling of control sequences and dynamics rollouts. Sampled trajectories are ranked using a game-aware cost to capture competitive interactions and generate diverse racing behaviors. Our planner then performs feasibility selection by explicitly enforcing track-boundary and dynamic collision-avoidance constraints, ensuring safe and reliable transitions between racing strategies. Extensive simulations on challenging tracks show that SGTP achieves a 95.24% win rate and a 99.35% task-completion ratio in highly interactive races, with a mean computational time of 0.095 s over multiple iterative solving steps. We also demonstrate the successful application of SGTP in large-scale scenarios with up to 10 agents. We release our code and provide an open-source benchmark of multi-agent autonomous racing algorithms to facilitate future research. Project page: https://sgtp-racing.github.io/.

Problem

• autonomous racing
• multi-vehicle interaction
• real-time planning
• strategic diversity
• computational efficiency

Innovation

• Sampling-based Planning
• Game-Theoretic Reasoning
• GPU-Accelerated Rollouts
• Multi-Vehicle Autonomous Racing
• Dynamic Collision Avoidance