Key points are not available for this paper at this time.
Tele-driving refers to a novel concept by which drivers can remotely operate vehicles (without being physically in the vehicle). By putting the human back in the loop, tele-driving has emerged recently as a more viable alternative to fully automated vehicles with ride-hailing (and other on-demand transportation-enabled services) being an important application. Because remote drivers can be operated as a shared resource (any driver can be assigned to any customer regardless of trip origin or destination), it may be possible for such services to deploy fewer drivers than vehicles without significantly reducing service quality. In this paper, we examine the extent to which this is possible. Using a spatial queueing model that captures the dynamics of both pickup and trip times, we show that the impact of reducing the number of drivers depends crucially on system workload relative to the number of vehicles. In particular, when workload is sufficiently high relative to the number of vehicles, we show that, perhaps surprisingly, reducing the number of drivers relative to the number of vehicles can actually improve service level (e.g., as measured by the amount of demand fulfilled in the case of impatient customers). Having fewer drivers than vehicles ensures that there are always idle vehicles; the fewer the drivers, the likelier it is for there to be more idle vehicles. Consequently, the fewer the drivers, the likelier it is for the pickup times to be shorter (making overall shorter service times likelier). The impact of shorter service time is particularly significant when the workload is high, and in this case, it is enough to overcome the loss in driver capacity. When workload is sufficiently low relative to the number of vehicles, we show that it is possible to significantly reduce the number of drivers without significantly reducing service level. In systems in which customers are patient and willing to queue up for the service, we show that reducing the number of drivers can also reduce delay, including stabilizing a system that may otherwise be unstable. In general, relative to a system in which the number of vehicles equals the number of drivers (as in a system with in-vehicle drivers), a system with remote drivers can result in savings in the number of drivers either without significantly degrading performance or actually improving performance. We discuss how these results can, in part, be explained by the interplay of two counteracting forces: (1) having fewer drivers increasing service rate and (2) having fewer drivers reducing the number of servers with the relative strength of these forces depending on system workload. This paper was accepted by Baris Ata, stochastic models and simulation. Funding: This work was supported by the US National Science Foundation Grant SCC-1831140, and the Guangdong (China) Provincial Key Laboratory of Mathematical Foundations for Artificial Intelligence 2023B1212010001. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.01687 .
Benjaafar et al. (Mon,) studied this question.