A Spatial Learning-Based Fault Tolerant Lateral Tracking Control for Autonomous Driving

In this article, a spatial learning-based fault-tolerant control strategy is proposed for precise lateral tracking of autonomous vehicles subject to dynamical uncertainties, external disturbances as well as actuator failures. In order to facilitate the controller design, the uncertain vehicle dynamics are firstly transformed into a parametric form in the space domain, where the system uncertainties are reorganized and combined into the parametric and input distribution uncertainties. Furthermore, considering the under-actuated property of the vehicle dynamics, a novel technique in dealing with the non-square input distribution matrix is employed, in which a pseudo-like inverse matrix and a robust term are introduced into the controller to compensate the mismatch between the number of inputs and outputs. Then the proposed spatial learning-based fault tolerant control algorithm is developed, which is equipped with two adaptive parametric updating laws to estimate the parametric uncertainties and the multiplicative actuator faults correspondingly. Consequently, the convergence of the control algorithm is analyzed rigorously under the framework of composite energy function. Case studies verify the feasibility and effectiveness of the proposed control scheme.