Multi-UAV Cooperative Trajectory Planning Method for Passive Positioning Tasks

To address the issue of autonomous unmanned aerial vehicles (UAVs) neglecting their mission importance due to environmental factors during positioning tasks, and to improve the accuracy of UAV positioning while reducing the Geometric Dilution of Precision (GDOP) caused by UAV formation and position, a trajectory planning method for autonomous UAVs aimed at passive positioning tasks has been proposed. This method utilizes the Deep Q-Network (DQN) algorithm from reinforcement learning, with formation translation and rotation as the action space, allowing the UAV to learn and autonomously plan its trajectory. In the entire framework, a trajectory planning method for autonomous UAVs aimed at passive positioning tasks has been put forward. This method leverages the decision-making capability of reinforcement learning, incorporating GDOP to consider the variation of positioning error in the calculation of rewards at each step, enabling the UAV to continuously consider the change in positioning error during flight. Additionally, due to constraints such as fuel consumption on UAV flight, an evaluation metric for UAV trajectory is proposed, leading to a trajectory with lower overall positioning error and shorter flight distance.