Multi-Objective Optimization for Robotaxi Dispatch With Safety-Carpooling Mode in Pandemic Era

Autonomous driving has been successfully implemented in such particular areas as logistics distribution centers, container terminals, and university campuses. Robotaxi represents one of its important applications. This work studies a robotaxi dispatch problem during the pandemic era. It aims to design a robotaxi dispatch approach according to a defined severity degree of the pandemic, which can decrease a virus infection rate by reducing contact among passengers. It develops a multi-objective optimization model to minimize travel cost of robotaxis, waiting time of both robotaxis and passengers, and contact among passengers. A two-stage nondominated sorting genetic algorithm (NSGA-TS) is proposed to solve the problem. Three operations are designed to generate new solutions, which can ensure its solution diversity and speed up its convergence. Its effectiveness is verified via its comparison with two popular multi-objective optimization algorithms, i.e., multi-objective evolutionary algorithm based on decomposition (MOEA/D) and nondominated sorting genetic algorithm II (NSGA-II). Experimental results show that the proposed model can effectively reduce travel cost and waiting time. Besides, it can reduce the virus infection rate by decreasing contact among passengers at different severity degrees of the pandemic. This work is conducive for our society to building intelligent transportation systems in the post-pandemic era.