A Model-Based Informed Eco-Routing Method for Electric Vehicles in Urban Environments

This paper presents a static route planning method for the energy-efficient autonomous driving of battery-powered electric vehicles in urban areas. Based on an analytic model of the vehicle energy consumption and a given road graph, a special A*-search algorithm is proposed to find the route that minimizes a weighted time-energy cost. For each road section, the proposed approach finds the solution to an optimal control problem constrained by traffic rules, road curvature and physical limits of the vehicle. This allows the method to be used with basic, widely available knowledge of the road network (road layout, infrastructure and speed limits). No learned or statistic traffic-related data like average speed or vehicle density is needed. Special attention is given to the effects of acceleration/braking profiles along the trajectory and to the energy-regenerating capacities of the electric vehicle, as we show they are major sources of energy consumption. Results show that our approach saves on average 53.3% on energy consumption compared to the fastest (safe and legal) driving, with only 6.8% longer travel time. It is also able to find routes that are on average 2.1% better than the shortest route when both routes are driven with optimal eco-driving. Our approach considers more energy-saving driving compared to other state-of-the-art techniques, mainly by featuring coasting. Thanks to a simple reformulation of the eco-routing cost, we propose the first informed graph search method for the eco-routing of vehicles with regenerative capabilities. We show on a practical example that this informed graph search saves can divide by 5 how much of the graph needs to be explored.