Model predictive control for vehicle steering using road information in the local frame

Abstract

This thesis was motivated by Chalmers formula student driverless 2019 (CFSD19) project, with the aim to deliver a self-driving formula race car and compete in the Formula Student Czech competition. In this thesis, a control strategy to control the lateral motion of the CFSD19 car was designed and simulated. The control algorithm is based on the model predictive control (MPC) framework. The simulation used a single track dynamic model (also known as the bicycle model) for simulating the car's motion, and a kinematic model, linearized around the currently sampled state at every sampling update, for the controller. The goal of the controller is to steer the car, so that it can track a given path in the global frame, using only track information observed by the car in its local frame. The performance of the controller is evaluated by measuring the lateral deviation of the car from the track. Simulation shows that the car is able to track the path with acceptable lateral deviation, and the control scheme can be executed fast enough for real-time application.