Real-Time Distributed Model Predictive Control with Gaussian Process Learning for Planar Vehicle Platooning

This paper proposes a distributed Model Predictive Control framework for planar vehicle platooning that integrates Gaussian Process Regression to compensate unmodeled dynamics and propagate uncertainty within the prediction horizon. The controller is developed consistently with a communication-aware supervisory philosophy and assessed under representative operating conditions associated with varying vehicle-to-vehicle communication quality. It is implemented as a real-time sparse sequential quadratic programming scheme and evaluated in multi-vehicle simulations, including emergency braking, lane changes, and communication-loss scenarios. Results show that a single controller core can ensure safe spacing, coordinated motion, and robust path tracking across diverse platooning conditions.