Abstract:In recent years, the visual navigation and autonomous flight of bionic flapping wing vehicles have become a promising research direction in the field of robotics, however, its limited carrying capacity puts forward strict restrictions on the type, size and weight of vision sensors. At present, the existing commercial image processing modules are difficult to meet the accuracy requirements under the limitation of size and weight, and need to transmit image information back to the ground control system for processing. This paper aims to design a lightweight airborne monocular vision system to help miniature bionic flapping wing aircraft obtain external information and realize intelligent and autonomous flight. Compared with other image processing modules, this system is only 2.7cm*2.5cm in size, weighs only 3g, and can complete image processing without image backhaul. Using the domestic high-computing power chip K210 as the main chip of image processing, the internal compatibility with lightweight network model to complete the real-time classification and recognition function, which through the serial port and the main control of the bionic aircraft for information interaction, can enable the bionic flapping wing aircraft to achieve gesture recognition and target tracking.