Path Tracking Using Linear Timevarying Model Predictive Control for Autonomous Vehicle

Abstract:

In order to improve realtime robustness of autonomous vehicle, a path tracking approach based on linear timevarying model predictive control is investigated. The longitudinal and lateral vehicle nonlinear dynamics model is established for verification of controller simulation. Starting from twowheeled model with 3 degrees of freedom (DOFs), linear timevarying path tracking predictive model is deduced. Vector relaxation factors are introduced to deal with the nonfeasible solution caused by the hard constraints in the optimization process. Based on model predictive control theory, the design of path tracking algorithm can be transformed into an online quadratic programming problem with soft constraints. Finally, both vehicle modeling and controller design are realized based on the Matlab/Simulink software, simulation results of double lane change show that the controller can adapt to robustness of different speeds and design parameters.