Online Ride-Hitching in UAV Travelling (2108.09606v1)

Abstract: The unmanned aerial vehicle (UAV) has emerged as a promising solution to provide delivery and other mobile services to customers rapidly, yet it drains its stored energy quickly when travelling on the way and (even if solar-powered) it takes time for charging power on the way before reaching the destination. To address this issue, existing works focus more on UAV's path planning with designated system vehicles providing charging service. However, in some emergency cases and rural areas where system vehicles are not available, public trucks can provide more feasible and cost-saving services and hence a silver lining. In this paper, we explore how a single UAV can save flying distance by exploiting public trucks, to minimize the travel time of the UAV. We give the first theoretical work studying online algorithms for the problem, which guarantees a worst-case performance. We first consider the offline problem knowing future truck trip information far ahead of time. By delicately transforming the problem into a graph satisfying both time and power constraints, we present a shortest-path algorithm that outputs the optimal solution of the problem. Then, we proceed to the online setting where trucks appear in real-time and only inform the UAV of their trip information some certain time $\Delta t$ beforehand. As a benchmark, we propose a well-constructed lower bound that an online algorithm could achieve. We propose an online algorithm MyopicHitching that greedily takes truck trips and an improved algorithm $\Delta t$-Adaptive that further tolerates a waiting time in taking a ride. Our theoretical analysis shows that $\Delta t$-Adaptive is asymptotically optimal in the sense that its ratio approaches the proposed lower bounds as $\Delta t$ increases.