LLA-MPC: Fast Adaptive Control for Autonomous Racing

2025-05-26

AI Summary

Real-time adaptive control of high-speed autonomous racing vehicles remains challenging under abrupt changes in tire-road friction coefficients. Method: This paper proposes a learning-free, zero-startup-delay online control framework centered on a “retrospective–prospective” dual-window mechanism: a retrospective window dynamically selects the most behaviorally matched dynamical model from a model bank, while a prospective window performs adaptive model predictive control (MPC) trajectory planning via receding-horizon optimization; friction coefficient estimation is integrated online without offline training or historical data accumulation. Contribution/Results: Experimental validation on multi-surface tracks demonstrates a 3.2× improvement in adaptation speed, significantly enhanced cornering stability, and robust high-precision trajectory tracking even under sudden low-adhesion conditions.

Technology Category

• Cognitive Modeling & Cognitive Systems: Adaptive Behavior
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Search and Optimization: Learning to Search

Application Category

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

Abstract

We present Look-Back and Look-Ahead Adaptive Model Predictive Control (LLA-MPC), a real-time adaptive control framework for autonomous racing that addresses the challenge of rapidly changing tire-surface interactions. Unlike existing approaches requiring substantial data collection or offline training, LLA-MPC employs a model bank for immediate adaptation without a learning period. It integrates two key mechanisms: a look-back window that evaluates recent vehicle behavior to select the most accurate model and a look-ahead horizon that optimizes trajectory planning based on the identified dynamics. The selected model and estimated friction coefficient are then incorporated into a trajectory planner to optimize reference paths in real-time. Experiments across diverse racing scenarios demonstrate that LLA-MPC outperforms state-of-the-art methods in adaptation speed and handling, even during sudden friction transitions. Its learning-free, computationally efficient design enables rapid adaptation, making it ideal for high-speed autonomous racing in multi-surface environments.

Problem

• Adapts to rapidly changing tire-surface interactions in autonomous racing
• Eliminates need for data collection or offline training periods
• Optimizes real-time trajectory planning for multi-surface racing scenarios

Innovation

• Real-time adaptive control for racing
• Model bank enables immediate adaptation
• Look-back and look-ahead mechanisms optimize trajectory