Research on flying-shooting method of robot based on KALMAN filtering and deep learning

Aiming at the problem that the traditional robot system for flying- shooting has the separation of vision and control, which leads to the low real-time and poor efficiency of the system, this paper puts forward the scheme of integration of vision and control, and studies the relevant algorithms of flying-shooting. In order to improve the real-time performance of the system, a hard real-time system based on windows is adopted as the operating system. The communication of robot control module and vision module is realized by sharing memory in the same operating system. Based on the improved KALMAN Filtering, a sensorless accurate time trigger algorithm is designed to reduce the positioning error of the robot at the same position when the camera is triggered. The processing speed of image is improved by deep learning, which ensures to complete the calculation of deviation value before the next beat of the robot, and finally the accurate positioning of flying- shooting is realized. The experimental results show that compared with the traditional separate scheme of vision-control system, the robot flying-shooting system for correcting deviation designed in this paper has the advantages of simplicity, precision and efficiency.