In our work, an algorithm is presented for tilt‐rotor UAV clusters in an unknown complex environment, which addresses the challenge of navigating and positioning in complex unknown environments without external navigation and positioning information. It introduces a deformable attention‐based UAV ground microtarget visual, LiDAR, and inertial navigation fusion‐tracking algorithm. This approach leverages an improved deep learning vision system to track ground microtargets and fuses the tracking results with LiDAR and inertial navigation data to achieve precise perception of both the tracking targets and the complex environment. Furthermore, we propose the algorithm for tilt‐rotor UAV cluster based on precise positioning and environment perception. Combining the environmental information, a spatial and temporal motion planning approach is designed, facilitating the trajectory planning of the tilt‐rotor UAV cluster. This ensures that the drones maintain a safe distance while achieving dense collaboration. And based on the structural characteristics of the tilt‐rotor UAV, an extended state observer is designed and used to accurately estimate composite interference. With our method, external disturbances in a complex environment can be accurately estimated, and feedforward suppression can be performed accordingly. As a result, the trajectory can be tracked in an accurate manner, thus improving the intelligence level of the drone cluster when executing tasks in an unknown interference environment.
Yang et al. (Thu,) studied this question.