Delay-Sensitive Goods Delivery and In-Situ Sensing Using a Multi-Task Drone

Drones are evolving into highly capable and adaptable devices, prompting the development of advanced control frameworks. This paper introduces a novel online control framework tailored for a multi-task drone, explicitly addressing the simultaneous execution of in-situ sensing and goods delivery. To...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on mobile computing Ročník 24; číslo 10; s. 10055 - 10068
Hlavní autoři: Liu, Bin, Ni, Wei, Liu, Ren Ping, Guo, Y. Jay, Zhu, Hongbo
Médium: Magazine Article
Jazyk:angličtina
Vydáno: IEEE 01.10.2025
Témata:
Complexity theory
delivery
Drone
Drones
Energy consumption
finite-horizon Markov decision process
Heuristic algorithms
in situ sensing
multitask
Multitasking
Navigation
Real-time systems
Schedules
Sensors
Trajectory
ISSN:1536-1233, 1558-0660
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Drones are evolving into highly capable and adaptable devices, prompting the development of advanced control frameworks. This paper introduces a novel online control framework tailored for a multi-task drone, explicitly addressing the simultaneous execution of in-situ sensing and goods delivery. To tackle this complex scenario, a finite-horizon Markov decision process (FH-MDP) is formulated to ensure not only the prompt delivery of goods but also the minimization of energy consumption and the maximization of the drone's reward for in-situ sensing. A significant contribution lies in establishing the monotonicity and subadditivity of the FH-MDP. This mathematical foundation provides evidence for the existence of an optimal, monotone, deterministic Markovian policy. The crux of the optimal policy revolves around flight distance- and time-related thresholds, determining the precise points at which the drone should switch its optimal action. This unique feature empowers the multi-task drone to make real-time decisions, such as adjusting flight speed or engaging in in-situ sensing, by comparing its current state with these predefined thresholds. This process can be accomplished with a linear complexity, ensuring efficiency in decision-making. The optimality of our approach is rigorously demonstrated through numerical validation, where it is compared against a computationally expensive, dynamic programming-based alternative. Under the considered simulation settings, our approach reduces drone energy consumption by a substantial 19.8% compared to existing benchmarks. This not only highlights the practical effectiveness of the proposed framework but also underscores its potential for significant advancements in the field of drone operations and energy efficiency.
ISSN:1536-1233
1558-0660
DOI:10.1109/TMC.2025.3570437