Robust lateral control for scaled autonomous vehicles using LPV polytopic approach
Résumé
This paper presents the design, analysis, and experimental validation of a robust lateral control system for a scaled autonomous vehicle. More precisely, this work introduces a Linear Parameter Varying (LPV) polytopic controller with a novel design adapted to manage the lateral dynamics of vehicles for path-following tasks, by the integration of dynamic look-ahead distance as an independent LPV scheduling variable and an augmented look-ahead error model. Key challenges include handling varying longitudinal velocities and maintaining good performance under diverse operating conditions. In addition, we introduce a geometry-based reference yaw rate generation algorithm, designed for discrete map data and optimized for realtime execution using KD-tree search. We evaluate the performance of the controller through experimental tests on a scaled vehicle. Experimental results demonstrate that the proposed approach achieves precise path following with lateral errors consistently below 6 cm under varying longitudinal velocities, confirming its stability and robustness under speed variations and bridging the gap between theoretical LPV control and practical implementation.