Cooperative Localization Method of UAVs for a Persistent Surveillance Task

The UAV(unmanned aerial vehicle) has outstanding movement ability, low price and convenient operation, which is suitable for carrying a specific camera to perform the persistent ground surveillance task. In the persistent surveillance task, the UAV locates the ground target according to the monocular localization method. In this paper, a cooperative localization method for persistent surveillance task is proposed to give full play to the advantages of multiple UAVs and deal with the target misdetection. Based on the monocular vision localization system, each UAV obtains its own localization results and broadcasts them. The cooperative method verifies the rationality of all localization results, scores all reasonable results through the error evaluation function, and obtains the result with the highest score as the localization result of the whole group. When there is no reasonable result, the method predicts the real-time location of the target according to the historical trajectory. Using the existing hardware platform, experiments are designed to verify the superiority, stability and expansibility of the method. Experimental results show that this method improves the localization accuracy of the UAVs, completely filters the abnormal data caused by target misdetection, and can be applied to other types of hardware platforms by modifying the sampling interval, lower distance limit and upper speed limit.