Iterative Learning Control for Lateral Tracking With Repeated Path in Autonomous Vehicles for Dynamic Environments

Many control techniques are proposed in literature for the lateral tracking of autonomous vehicles. Unfortunately, most of these demonstrate control efficiency based on supportive results from simulation only. A lack of evidence in experimental verification raises questions about the design performance. This research proposes a practical tracking algorithm using iterative learning control (ILC) with the expectation of achieving a lower tracking error in each iteration. ILC observes an error sequence from the past to adjust the steering command in a real vehicle. This work also develops a localization system to achieve the real-time positioning data based on extended Kalman filtering (EKF). Signals obtained from the inertial measurement unit (IMU), global navigation satellite system (GNSS) module, and vehicle encoders are fused to determine the real-time vehicle position. Experiments were conducted to validate the effectiveness of the ILC-based tracking control. The results show that the ILC designs can clearly improve the tracking performance from a typical control system by reducing the tracking error in the iteration domain. In addition, using more gains in the ILC design results in a smoother path.