Automotive Model-Based Differential System Development for Steering Control of an Autonomous Vehicle

This thesis is related to the almost new context of fully-autonomous vehicles, focusing on the problem of steering optimization for a two-wheel drive (2WD) electrical vehicle, employing an algorithm that reproduces a differential gear. The vehicle is scaled form real with a 1:32 factor and fully-autonomous. Using a line scan camera, the border lines of the track are recognized, then a on-board Control Unit (CU) elaborates the informations coming from the camera and provides command signals to the actuators, which keeps the vehicle in the track limits. Trajectories are not stored, but elaborated in real-time. The objective of the thesis was to obtain a block in model-based software, executing an algorithm, which computes the speeds references for the posterior wheels. In order to realize the differential algorithm, the kinematic of the vehicle has been used. Then, from steering angles, algorithm equations have been computed, using the inverse kinematic. These equations allow the computation of speeds ideally needed by posterior wheels, in order to steer with constant curvature radius. To ensure speeds required, was needed a speed control for the DC motor, that needed a speed sensor implementation. Two typologies of sensors have been produced, those differ for the logic of speed computation and they implementation that has been developed, using language codes specific for the model-based software. The control realization was based on the system model but, in the case of study, the system parameters were unknown. For that reason an identification method has been applied in order to obtain a system model and control it. Afterwards the differential algorithm has been tested and verified under two different situations: with steering servo management and without it.