Robot Routing Problem of Last-Mile Delivery in Indoor Environments

Abstract

With the development of robot technology, trials adopting robots for last-mile delivery are continuing, and the final destination of last-mile delivery is further expanding into indoor environments. Unlike existing studies conducted for robot-based last-mile delivery in outdoor environments, two main issues must be solved to enable last-mile delivery in indoor environments using robots. First, it is necessary to reasonably and realistically estimate the robot travel time considering horizontal and vertical movement segments within a given building. Second, optimizing the robot routing problem based on the estimated robot travel time is necessary. In this paper, we proposed a new method to estimate the robot travel time considering robot movement characteristics and an elevator in a building. In addition, we developed a mathematical model of the robot routing problem and problem-specific heuristic based on a genetic algorithm to quickly solve the proposed mathematical model. It obtained the exact solutions when the problem size was small and near-optimal solutions in the medium- and large-sized problems (average optimality gap: 0.11% and 0.18%, respectively). Through extensive experiments assuming various building structures, it was determined that the proposed model and heuristic can quickly yield realistic solutions for indoor robot-based last-mile delivery.