An Intelligent Emotional Real-Time Learning Controller for High-Speed Lane Keeping Assist System in Autonomous Vehicles

This article presents a novel intelligent control algorithm for the Lane Keeping Assist System based on Emotional Learning. The proposed controller can be trained online using the continuous stress signal generated by the critic unit during the control process. Fast learning convergence, stable behaviour, and high adaptability and robustness against disturbances are the main advantages of this controller when compared to other controllers such as PID, LQR, and MPC. The performance of the controller is validated in a high-speed lane following task on a winding road. It is shown the proposed controller performed better in both initial settlement and disturbance rejections with up to 50% less tracking error in the presence of high lateral wind force. Stability and adaptability to the vehicle's dynamics are evaluated with different velocities and surface friction which ended up with 30% less tracking error in lateral position for slippery surfaces (25% less friction). Moreover, the computation cost is 80% lower compared to MPC. In contrast to other reinforcement learning algorithms, the proposed controller does not require a high training effort which makes it suitable for use in real-time applications.