Distributed Leader-Follower Model Predictive Control for Multi Platoon Coordination under Signal Temporal Logic specifications

Sammanfattning

In this work, a hierarchical model predictive control approach for the coordination of multiple platoons is proposed. Today a majority of transportation is still happening on the road. This leads to congested roads, greenhouse gas emissions, and accidents. A platoon is a group of vehicles moving in a string but without any physical coupling. Instead, the vehicles are coupled by inter-vehicle communication that allows them to exchange information with one another. Platooning leads to several benefits: The exchange of information can allow the vehicles to reduce the distance between them without decreasing safety and further allow vehicles to accelerate and decelerate less and hence increasing the traffic flow. A better traffic flow leads to more safety, less congestion, and less fuel consumption and therefore improving all issues caused by road transportation. The control problem is distributed into a control problem for the platoon leaders and a control problem for the platoon followers. While the leader control problem is independent of the follower’s trajectories, the follower control problem depends on the leader’s trajectories. The two controllers are therefore structured hierarchically. Furthermore, the leader’s control problem includes Signal Temporal logic specifications. The parameters for these specifications are obtained by solving an offline optimization problem. The proposed controller framework is therefore structured in three levels. On the highest level, the coordination of multiple platoons is planned by determining time instances at which the platoons should reach specific road segments and further determining reference velocities for the platoons to achieve this. On the middle level, platoon leaders follow the obtained reference velocities, but also consider safety issues that can occur, using Signal Temporal Logic encoded as Control Barrier Functions in a Model Predictive Controller. The Signal Temporal Logic specifications use the time parameters obtained at the top level. On the lowest level, another Model Predictive Controller organizes the follower vehicles within a platoon and ensures that they follow the same trajectory without the distance between them becoming too small or too large. The performance of the proposed controller framework is evaluated in two simulation scenarios including merging maneuvers and constraints on some road segments. Furthermore, recursive feasibility for all optimization problems is shown.