An Efficient Path Planning of Multiple UAVs Using Novel Chaotic Frilled Lizard with Deep Reinforcement Learning Model

Multi-UAV planning is a technique for scheduling several Unmanned Aerial Vehicles (UAVs) to move over an area effectively while avoiding obstacles and achieving targets, including traveling distance, mission completion time, and energy consumption. Planning is necessary for many applications such as delivery services, search and rescue, and surveillance. The primary goal is to find accident-free routes that allow UAVs to reach their targets as quickly and with the least amount of resource usage possible. Several techniques have been developed for multi-UAV path planning. However, some drawbacks were identified, such as lower throughput and energy efficiency, higher delay rate, and so on. Thus, the proposed technique is introduced to address these issues and enhance the performance of multi-UAV path planning. A novel Chaotic Frilled Lizard with Deep Reinforcement Learning (CFL-RDL) model is created to facilitate the path planning process of multiple UAVs. The most efficient UAV routes are selected using chaotic frilled lizard optimization, and reinforcement learning is used to balance the optimizer’s exploration and exploitation phases, allowing for autonomous decision-making throughout the operation. The efficiency of the proposed technique is determined by comparing it with various existing techniques. In this paper, path planning has been performed for 10 UAVs. The proposed technique attains 2.335 bits throughput for UAV1, 2.993 bits throughput for UAV2, 2.704 bits throughput for UAV3, 2.586 bits throughput for UAV4, 3.086 bits throughput for UAV5, 3.239 bits throughput for UAV6, 2.522 bits throughput for UAV7, 2.903 bits throughput for UAV8, 3.436 bits for throughput UAV9 and 3.012 bits throughput for UAV10.