Revisiting multi-objective three-dimensional trajectory planning in dynamic unmanned aerial vehicles assisted systems using online Monte Carlo Search methods

Trajectory planning for Unmanned Aerial Vehicles (UAVs) in complex three-dimensional (3D) environments is a critical challenge in UAV-assisted task planning systems, particularly in dynamic scenarios such as rescue operations where UAVs serve as mobile wireless communication hubs. This paper addresses this challenge by revisiting Monte Carlo Search methods and proposing a novel path planning framework with a heuristic-based Nested Monte Carlo Search (NMCS) algorithm, designed to optimize the UAV’s trajectory in real-time while supporting a rescue team with computational task offloading. To simulate the dynamic movements of the rescue team, we employ a Gauss-Markov Mobility Model (GMM), which enhances the realism and adaptability of the environment. Unlike Deep Reinforcement Learning (DRL) methods, which often require extensive offline training and struggle with real-time adaptability, NMCS operates as an online search algorithm, offering superior responsiveness to dynamic environmental changes. Our proposed NMCS framework utilizes a two-level nested architecture to effectively balance three competing objectives: maximizing communication throughput, minimizing flight path length, and reducing energy consumption. Furthermore, the algorithm demonstrates robust performance in maintaining fairness among multiple rescue teams, ensuring equitable resource allocation. Extensive simulations validate the effectiveness of our approach, showing that the NMCS-based algorithm significantly outperforms benchmark methods in handling complex, dynamic, multi-objective 3D path planning problems. This work highlights the potential of NMCS as a powerful tool for UAV-assisted wireless communication systems, particularly in dynamic and time-sensitive operational scenarios.