Multiple Unmanned Aerial Vehicle Perception Method Based on Radar/Infrared Information Fusion

Target perception with Unmanned Aerial Vehicles (UAV) is expected to expand the detection space in complex environments and realize high-dimensional situational cognition. Aiming at the problems of inefficient multisource information fusion and poor accuracy of maneuvering target tracking, a Multiple UAV Cooperative Perception Method based on Radar/Infrared information fusion (MCPMRI) is proposed. In this method, the previous moment trajectory information is disassembled into the particle cloud, which is then corrected using radar/infrared observations. Hierarchical clustering is employed to eliminate jamming, information entropy is used to assign weights, and fusion resampling is conducted to arrive at the current estimated state. Unlike traditional methods, MCPMRI corrects the particle weights based on the probability density of sensor observations, utilizes the particle discrete variance matrix determined by the state at previous moments, performs target association based on hierarchical clustering, and assigns the estimated state weights of each UAV based on information entropy. This approach enhances the adaptive capture of maneuvering targets, transmits uncertainty information, and systematically eliminates pseudo-targets, thereby improving estimation accuracy and reliability. Simulation results show that with MCPMRI, the UAV formation equipped with radar/infrared sensors achieves pseudo-target recognition, antijamming tracking, and higher target tracking accuracy compared to traditional algorithms.