Gain-scheduled LPV Lateral Control for Lane Tracking and Lane Change Maneuvers for Autonomous Vehicles
Résumé
This paper presents an integrated Linear Parameter Varying control (LPV) structure capable of handling both lanetracking and lane-change maneuvers for autonomous vehicles. Within the LPV framework, the entire operating domain of velocities is considered. Two distinct look-ahead time profiles have been selected for lane-change and lane-following scenarios. By introducing an exogenous varying parameter to the look-ahead error equations, an LPV Gain-scheduled controller has been designed using H∞ parameter-dependent weighting functions to simultaneously address both control problems. For the two distinct cases of lateral control, the LPV controller is implemented in real-time as an interpolation of the gridded state space controllers according to the induced parameter, which is a function of the lateral deviation of the car. Consequently, the novel Gain-scheduled controller, at every instant, depends on the two different look-ahead time profiles according to the vehicle's velocity and the on-line computation of the induced parameter. The novel lateral control algorithm undergoes validation, yielding promising outcomes in a high-reliability simulator capable of handling both single and double lane-changes. Subsequently, experimental tests are conducted, employing an automated electric Renault Zoe as a test bed platform, showcasing the effective performance of the proposed control system in realworld scenarios.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |