A real-time multiobjective optimization algorithm for discovering driving strategies

Abstract

Vehicle driving consists of selecting and applying the best control actions in real time to optimize several objectives such as the traveling time and the fuel consumption. Because more than one objective is optimized, this problem can be solved using multiobjective optimization techniques. However, the existing optimization algorithms mostly combine objectives into a weighted-sum cost function and solve the corresponding singleobjective problem. To test the multiobjective approach, we developed the multiobjective optimization algorithm for discovering driving strategies (MODS) that searches for the best driving strategies by taking into account the entire route. Although this algorithm, on average, outperforms existing single-objective algorithms for discovering driving strategies, it has a drawback, namely, it cannot be used for real-time optimization because of its time complexity. To overcome this shortage, we redesigned the MODS algorithm, obtaining the real-time multiobjective optimization algorithm for discovering driving strategies (MODS-RT). The MODS-RT algorithm was tested on data from real-world routes and compared with MODS and traditional single-objective algorithms for discovering driving strategies. Although MODS-RT found worse driving strategies than MODS, it found better driving strategies than the single-objective algorithms, thus proving that the multiobjective approach can be effectively adapted for real-time discovery of driving strategies.