An Integrated Framework for Planning and Control of Semi-Autonomous Vehicles

This thesis presents the design of a novel activesafety system preventing unintended roadway departures. Theproposed framework unifies threat assessment, stability, andcontrol of passenger vehicles into a single combined optimizationproblem. A nonlinear Model Predictive Control (NMPC) problemis formulated where the nonlinear vehicle dynamics, in closed-loop with a driver model, is used to optimize the steering andbraking actions needed to keep the driver safe. A model of thedriver's nominal behavior is estimated based on his observedbehavior. The driver commands the vehicle while the safetysystem corrects the driver's steering and braking action in casethere's a risk that the vehicle will unintentionally depart theroad. The resulting predictive controller is always active andmode switching is not necessary. We show simulation resultsdetailing the behavior of the proposed controller as well asexperimental results obtained by implementing the proposedframework on embedded hardware in a passenger vehicle. Theresults demonstrate the capability of the proposed controller todetect and avoid roadway departures while avoiding unnecessaryinterventions