Driving from afar: The vehicle routing problem with remote drivers
研究了远程司机可在服务点换车的车辆路径问题，建模为混合整数规划并用Benders分解求解，发现远程驾驶能有效减少司机需求。
Autonomous driving is about to change the transportation of people and goods. However, due to very high safety requirements, even in exceptional situations, autonomous driving may take longer than anticipated to reach the necessary technological maturity and finally be approved for use on public roads. Remote driving, in which humans in the loop teleoperate vehicles, can be a key bridging technology. Even when autonomous driving is the standard, remote human supervisors will remain crucial for solving problems in exceptional situations. In this context, this paper extends the classical vehicle routing problem with time windows (VRPTW) by the specificities of remote driving. Remote drivers need not remain in their vehicles during service times at customer locations. Instead, they can switch to other vehicles, opening the possibility of operating a fleet with fewer drivers. We model a basic version of the resulting VRPTW with remote drivers (VRPTWRD) as a mixed integer program and resort to Benders decomposition to derive effective branch-and-Benders-cut approaches. Our computational experiments analyze both algorithmic performance and managerial insights, including changes in routing costs, driver utilization measured through idleness rates, and operational flexibility quantified by driver switches. The results reveal that remote driving can be an effective lever to reduce the workforce in times of human driver shortages.