UAV Swarm Control Algorithm Based on Soft Attention Mechanism

Aiming at the problem that the average mechanism in the current UAV swarm interaction process is not timely enough to respond to the individual emergency response, a neighbor set interaction model based on soft attention mechanism is proposed. The neighbor set is updated according to the self information content contained in the neighbor, so as to improve the response to the individual emergency operation in the swarm, which is conducive to the realization of obstacle avoidance, turning and other functions of the swarm. On the other hand, the task is decoupled from the swarm control, and a swarm control method based on the virtual long machine mechanism is established. The virtual long machine is added to the UAV neighbor swarm, and the movement of the UAV swarm is controlled through the interaction between the UAV and the neighbor.