Fleet Speed Profile Optimization for Autonomous and Connected Vehicles

A fuel cost optimization method for a fleet of autonomous electric vehicles is proposed using the total cost of the fleet to generate the optimum speed profile. The proposed method also assures maintaining a safe distance from the adjacent vehicles and safe lane changing on multilane roads. While the method aims to get the optimum fleet speed, the individual speed of each vehicle is adjusted based on the relative distance from the leading vehicle. A deep neural network (DNN) based autonomous electric vehicle modeling considering all the route conditions is provided. Extensive simulations and experimental studies were performed to develop vehicle models and individual signatures to include in the process. The optimization is performed for a fleet of multiple vehicle types where the physical models of the vehicles vary quite a bit. The proposed optimization algorithm was implemented for different case studies, and it shows a consistent reduction of the total fuel cost for the fleet compared to the optimization done based on only the leading vehicle's cost and a significant improvement compared to unoptimized fleet operation.