Time domain disturbance observer based steering angle control for autonomous path following

Abstract

Path following in autonomous vehicles consists of tracking a given path with minimum deviation by setting inputs to the system's actuators. Several path tracking controllers are proposed in literature, and several of those offer accurate tracking performance in normal driving conditions. Road and vehicle characteristics are known to be widely variable and those are difficult to estimate in real time. Disturbance observers (DOB), when applied to a given system, are known to approximate the system response to that of its model by estimating and compensating for parameter changes, modeling and parameter identification errors and external disturbances. Common path following control laws, namely state feedback control and Stanley are evaluated in association with DOB for steering control of autonomous path following. The performance of these algorithms are compared to that of model-based predictive control (MPC) in the presence of disturbances. Simulation and experimental results are used in order to illustrate the performance gains associated to the usage of disturbance observers in such context.