Eco-Driving of Connected Autonomous Vehicles at Railway Level Crossings

Eco-driving can broadly be defined as a combination of decisions that can be made to increase energy efficiency of the vehicle. This is a complex decision making process involving many considerations, including the vehicle's states and its dynamic environment. At railway level crossings, it becomes even more complex because drivers must stop to check for approaching trains, causing long queues and traffic delays. In this thesis, we design and develop an automated driving system that coordinates between vehicles to eliminate queueing and improve fuel economy and road capacity. We do this through mathematical optimization models built using real driving data collected at Shin-Kiryu Railway Station in Gunma, Japan. Our results show improved fuel economy, idling times, and reduced queueing. We also recommend a policy to waive the mandatory stop regulation at Japanese railway level crossings in future autonomous-dominated futures.