Mixed feature visual servo method combined with reinforcement learning

The invention relates to a mixed feature visual servo method combined with reinforcement learning, and belongs to the technical field of servo control. Internal and external parameters of a system are obtained through a calibration board, then feature information of an image is obtained through operation such as image filtering, point light flow is tracked through a KLT algorithm, moment information of the image is obtained through global information of the image, then driving debugging of the mechanical arm is completed, and operation of the whole system is prepared. According to the method, respective optimization of visual servo of point features and image moment features is combined, so that the methods of the point features and the image moment features are complementary. And point feature tracking is carried out by adopting KTL, so that the method does not depend on a special target image. And reinforcement learning is added, and the control output of the system is interfered through the decision-making