Autonomous Refueling Strategies using Vehicle-to-Infrastructure Communication in Smart Cities

Recently self-driving vehicles have attracted tremendous attention from all walks of life. A problem facing self-driving vehicles is when to stop for gas. In this paper, we study the autonomous refueling strategies using vehicle-to-infrastructure communication in smart cities. We set three goals for our strategies: not to stop too late, nor too early, and get relatively cheap gas. To satisfy these goals, we relate our problem to the Gusein-Zade's version of the secretary problem and provide a solution framework where we divide the distance a vehicle can travel from a full tank to an empty tank into Density Observation Section, Secretary Section, and the Critical Section. To predict the number of gas stations a vehicle will encounter in the future, we use the Constant Density Approximation (CDA) method and the Machine Learning (ML) method. Simulation results comparing the proposed CDA and ML algorithms with the ground truth algorithm show that both perform nearly as well as the ground truth in satisfying all three goals.