Hierarchical Model Predictive Control Approaches for Strategic Platoon Engagement of Heavy-Duty Trucks

For a group of vehicles, collaborative platooning can be valuable in certain situations due to aerodynamic drag reduction, while being detrimental or altogether impractical in others. This paper details a platoon engagement/disengagement controller, which alternates between velocity trajectory optimization (VTO) in isolation and a fused platooning and VTO approach, capable of disengaging a platoon during segments detrimental to fuel savings and rejoining the platoon afterwards without significant energy expenditure. The proposed approach leverages parallel model predictive control (MPC) computations that (i) can identify when a platoon should be engaged/disengaged and (ii) performs the engagement/disengagement in a fuel-optimal manner. Using a medium-fidelity Simulink model furnished by Volvo, two real-world trucking routes, and two different traffic scenarios, the effectiveness of the approach was compared against non-platooning VTO, as well as a baseline controller that uses a PI-based cruise controller that incorporates Gipps car-following model. Results for a two-vehicle platoon using a 1 vehicle following distance reveal a 9.6% to 11.9% decrease in aggregate fuel consumption for both vehicles within the platoon, as compared to the baseline, highlighting the ability to disengage and rejoin a platoon without expending unnecessary fuel consumption. Additionally, the approaches for disengaging a platoon result in a 4-7% decrease in aggregate fuel consumption, as compared to a VTO-only approach.