Efficient multi-UAV coordination via rl: joint path planning and task scheduling for post-disaster UAV-assisted MEC systems

Unmanned aerial vehicles have emerged as an important technology in a wide spectrum of domains, encompassing environmental surveillance, precision agriculture, and disaster management, while serving critical functions in time-sensitive applications such as search and rescue operations in post-disaster scenario. However, existing studies on UAV trajectory design often treat path planning and task scheduling separately, assume static environments, or fail to address real-time dynamics. These issues create a gap in delivering efficient and adaptive UAV-assisted mobile edge computing (MEC) services under highly uncertain post-disaster conditions. In this paper, we propose the Geometric Reinforcement Learning Algorithm (GRLA), a unified framework for joint path planning and task scheduling in multi-UAV MEC systems. GRLA employs a unified reward matrix that integrates geometric distance, dynamic obstacle risk, and ground-user demand. At the same time, adaptive parameters allow flexible trade-offs between service quality and risk tolerance. Meanwhile, task scheduling is optimized using a greedy algorithm to reduce computational complexity, and UAVs cooperate through information sharing to achieve robust performance. Simulation results show that GRLA achieves a 100 % success rate, lower runtime, and improved quality-of-service (QoS) compared to GA, PSO, and RTT baselines. These findings highlight GRLA's robustness and efficiency, validating it as a practical solution for real-world post-disaster UAV coordination. The key contribution of this paper is the development of the GRLA framework, which unifies path planning and task scheduling in UAV-assisted MEC systems under dynamic post-disaster conditions.

Text 122259.pdf - Published Version Restricted to Repository staff only Download (14MB) Text 122259-accepted.pdf - Accepted Version Restricted to Repository staff only Download (3MB) Official URL or Download Paper: https://linkinghub.elsevier.com/retrieve/pii/S1389...

Actions (login required)

View Item