Lane Planning of Connected Autonomous Vehicles using a Graph Representation of Road Contents

Integrating Connected Autonomous Vehicles (CAVs) on the road presents numerous opportunities for safety and mobility enhancement. One of the main features that CAVs introduce is gathering data that conventional vehicles cannot currently access. Information such as traffic jams or hazards downstream can impact a driver’s lane choice, thereby improving or potentially worsening their experience. This information can be collected by CAVs and disseminated through a cloud system to vehicles upstream. Hence, significantly enhancing the lane-level decision making process. In this work, we propose a dynamic lane-level forward graph that considers different contents on the road that can be static, semi-static or dynamic. Our proposed approach effectively integrates map data as well as downstream data collected by CAVs to present a policy to optimally select the best lane given this information. To evaluate the proposed approach, simulations were carried out using the microsimulation tool, Aimsun, across different CAV market penetration rates. Results show that the proposed approach can reduce travel time by 20%, while also improving traffic safety.