Optimization of Robotaxi Dispatch With Pick-Up/Drop-Off-Point and Boarding-Time Recommendation

摘要

With the advancement of vehicular automation and communication technology, autonomous driving has emerged as a significant trend in future transportation. Robotaxis, an innovative mode of transportation that integrates robotics and artificial intelligence, are anticipated to become widely used, thereby revolutionizing urban mobility. This work proposes a multi-objective mixed integer programming model for robotaxi dispatch. Unlike previous approaches, it can recommend passengers’ pick-up points, drop-off points, and boarding time (BT) that may deviate from their initial origins, destinations, and BT, respectively. It encourages passengers to accept the recommended pick-up and drop-off (UO) points or to be picked up slightly earlier or later. The objectives are to maximize the profit per kilometer of robotaxis and to minimize the total travel expense of passengers. Subsequently, a nondominated sorting genetic algorithm with mass center (NSGA-MC) is proposed to solve the model. It outperforms nondominated sorting genetic algorithm II (NSGA-II) and multi-objective evolutionary algorithm based on decomposition (MOEA/D) across several metrics. Some instances provide detailed results that illustrate the effectiveness of the proposed algorithm. Additionally, a sensitivity analysis is performed, comparing scenarios with and without UO-point and boarding-time recommendations. An experiment is also conducted to examine various recommendation acceptance rates, demonstrating the advantages of the proposed model in developing intelligent public transportation systems.