Distributed optimization based algorithms for vehicle platooning

Abstract

The rise of vehicle usage causes roads to reach capacity limits. When capacity is reached, traffic jams and accidents are more likely to occur. By raising the efficiency of traffic, the capacity of roads can benefit without the need to expand infrastructure. Automating longitudinal controls of vehicles and creating vehicle platoons show potential to raise traffic efficiency. A platoon consists of multiple vehicles travelling closely behind each other. Automating the longitudinal controls shows promise to decrease distances between these vehicles, while maintaining safety and raising traffic throughput. This smaller distance is below that of conventional human control, thus improving traffic density. Most modern cars already have the option of being equipped with longitudinal control. The most advanced being Adaptive Cruise Control (ACC) which can either hold a predefined velocity or adapt that velocity to maintain distance to a preceding vehicle. However, due to the limitation of using only on-board sensors, these systems do not guarantee safety or increase traffic throughput. Research is being done on expanding ACC with Vehicle to Vehicle (V2V) communication creating Cooperative Adaptive Cruise Control (CACC). The addition of communication provides more precise data of neighbouring vehicles than conventional on-board sensing can provide and allows for vehicles to communicate intent. Optimization control for CACC systems has been researched, but it rarely incorporates passengers comfort or the future intent of the preceding vehicles. The goal of this thesis is to implement a decentralized optimization algorithm on a real-time simulated platoon of three vehicles, which takes into consideration safety, propagation of errors through the platoon, and comfort. These three aspects are subsequently implemented and evaluated on a real-time simulation platform in a distributed fashion.