Development of AI-based and control-based systems for safe and efficient operations of connected and autonomous vehicles

This research is in three parts. The first part recognizes the range limitations of onboard sensors such as LiDAR and cameras, and develops an AI control system that fuses sensed (local) information and longer-range information to make CAV lane-changing decisions. Deep Reinforcement Learning is being used to provide an end-to-end framework that will help identify the optimal connectivity range for each domain of prevailing operating traffic density. The second part is developing a method to demonstrate a CAV’s catalytic efficacy for addressing stop-and-go traffic perturbations that adversely affects operational efficiency, fuel economy, emissions, travel time, and driver/passenger comfort. The third part is developing a collision avoidance framework for CAVs, to reduce the likelihood of collision with surrounding vehicles, particularly HDVs that drive aggressively or have uncertain or unpredictable behavior.