Iterative Learning Distributed Model Predictive Control for Autonomous Vehicle Platoons With Applications to Repetitive Tasks

Autonomous vehicle platoons are particularly suitable for repetitive tasks due to their capability for efficient coordination, enhanced safety, and reduced driver fatigue. The offline-designed control policy faces difficulties in adapting to changing conditions without driver intervention or vehicle self-learning, which can lead to inadequate coordination among vehicles and an increased risk of collisions or disruptions. This paper presents an iterative learning distributed model predictive control (ILDMPC) strategy designed for 2-dimensional (2-D) autonomous vehicle platoons, allowing vehicles to learn from their previous iterations to minimize control errors and improve overall performance. First, the combined lateral and longitudinal dynamics incorporating load transfer of heterogeneous autonomous vehicle platoons are modeled together. Then, traffic regulations and mechanical constraints are defined and integrated into an optimal control problem with multiple objectives using the ILDMPC framework. This approach differentiates the platoon leader (PL) from the platoon followers (PFs) by employing distinct References. Additionally, the iterative learning is integrated into the optimal control problem via a convex terminal cost within a finite time horizon, completing the ILDMPC strategy. This strategy allows autonomous vehicle platoons to iteratively perform repetitive tasks, achieving optimal performance through iterative online learning. Simulations are carried out to demonstrate the effectiveness of the proposed controller, which is validated to evolve existing control laws and result in a 40% improvement as quantified by the error-based indicator. The associated codes are accessible at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/ZNianHua/ILDMPC</uri>