Multi-view collaborative tracking method and apparatus for fast moving target under low luminance condition

A multi-view collaborative tracking method for a fast moving target under a low luminance condition, wherein the method comprises a multi-view tracking algorithm for a fast moving target under a low luminance condition and an unmanned aerial vehicle (UAV) cluster collaborative tracking algorithm;

• the multi-view tracking algorithm for a fast moving target under a low luminance condition comprises the following steps:
• performing multi-source environment sensing by using sensors that comprise a millimeter wave radar sensor, a red green blue (RGB) visual sensor, and an infrared visual sensor and are carried by a UAV;
• normalizing, in a time dimension and a space dimension by using a three-space-time heterogeneous alignment algorithm, multi-source sensing data received by different sensors;
• inputting multi-source fused sensing data obtained through space-time normalization into a multi-source fusion sensing feature calculation network and calculating world coordinates of a tracked target under a current frame;
• and
• comparing a calculation result with a calculation result of a previous frame, and determining changes of a position and a moving direction of the tracked target in a world coordinate system;
• and
• the UAV cluster collaborative tracking algorithm comprises the following steps:
• sharing world coordinate information of the tracked target through cluster communication, and formulating and updating a trajectory of a UAV cluster through multi-agent reinforcement learning, to ensure that the tracked target is at a center of a sensing range of the UAV, and implement collaborative tracking of the UAV cluster.