QuayPoints: A Reasoning Framework to Bridge the Information Gap Between Global and Local Planning in Autonomous Racing

2025-10-12

AI Summary

In autonomous racing, a disconnect between global and local planners leads to suboptimal local decisions lacking time-optimal guidance. Method: This paper proposes QuayPoints—a framework that identifies critical regions along a globally time-optimal trajectory and explicitly encodes the penalty for deviating from it, yielding structured, semantic markers (QuayPoints) injected as contextual priors into the local planner. Contribution/Results: QuayPoints enables lightweight, interpretable, and computationally efficient transfer of global optimality knowledge to the local level. Experiments across four diverse race tracks demonstrate that integrating QuayPoints consistently enables the planning system to outperform opponents traveling at 75% of the ego-vehicle’s maximum speed. Moreover, it significantly improves the real-time responsiveness and robustness of strategic maneuvers—particularly overtaking—under dynamic racing conditions.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Multiagent Systems: Multiagent Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Autonomous racing requires tight integration between perception, planning and control to minimize latency as well as timely decision making. A standard autonomy pipeline comprising a global planner, local planner, and controller loses information as the higher-level racing context is sequentially propagated downstream into specific task-oriented context. In particular, the global planner's understanding of optimality is typically reduced to a sparse set of waypoints, leaving the local planner to make reactive decisions with limited context. This paper investigates whether additional global insights, specifically time-optimality information, can be meaningfully passed to the local planner to improve downstream decisions. We introduce a framework that preserves essential global knowledge and conveys it to the local planner through QuayPoints regions where deviations from the optimal raceline result in significant compromises to optimality. QuayPoints enable local planners to make more informed global decisions when deviating from the raceline, such as during strategic overtaking. To demonstrate this, we integrate QuayPoints into an existing planner and show that it consistently overtakes opponents traveling at up to 75% of the ego vehicle's speed across four distinct race tracks.

Problem

• Bridging information gap between global and local planning
• Preserving global time-optimality insights for local decisions
• Enabling informed deviations from optimal raceline during overtaking

Innovation

• Framework preserves global knowledge for local planning
• QuayPoints convey time-optimality information to planners
• Enables informed deviation decisions during strategic overtaking