Soutenance de these

Title: Advanced Estimation Algorithm in the Application to Autonomous Vehicles

Summary
This thesis develops observer-based state estimation for individual and platooned autonomous vehicles, addressing nonlinear dynamics, limited sensing, sensor faults, and V2V delays. It makes three main contributions:
Individual vehicle safety: An H∞ observer uses generalized-inverse techniques and relaxed rank conditions to handle nonlinear generalized systems. It estimates sideslip angle, rollover index, and unknown tire forces using only three conventional sensor signals.
Resilient platoon estimation and control: Distributed unknown-input observers correct sensor faults, while high-gain observers handle nonlinear dynamics. Observer-controller co-design reduces reliance on predefined spacing and shared control inputs while guaranteeing string stability.
CACC under unknown V2V delays: A nonlinear observer reconstructs preceding vehicle states from delayed V2V data. It ensures robustness to delay-estimation errors, balances delay tolerance and estimation performance, and supports false-timestamp detection.
CarSim simulations and Quanser QCar 2/QLabs experiments validate the proposed methods for rollover prevention, distributed estimation, and cooperative platoon control.
Members of the Jury
Reviewers
Enrica Zereik, Senior Research, Italian National Research Council (CNR), Italy
Ichalal Dalil, Full Professor, Paris-Saclay University, France
Examiners
Taous-Meriem Laleg-Kirati, Senior Research, Inria Paris-Saclay, France
Zehor Belkhatir, Lecturer, University of Southampton, United Kingdom
Fanwei Meng, Full Professor, Northeastern University at Qinhuangdao, China
Thesis Supervisors
Ali Zemouche, Full Professor, Université de Lorraine (Thesis Supervisor)
Marouane Alma, Associate Professor, HDR, Université de Lorraine (Co-supervisor)