Cardiovascular interventional surgery robot is an effective means to solve the problem of X-ray radiation injury and physiological fatigue of surgeons. With the development of robotics, robotic autonomous surgery has become one of the research hotspots. However, the current intelligence level of cardiovascular interventional surgery robots is low. Aiming at the guidewire delivery process of the designed cardiovascular interventional surgery robot, this paper proposes a new control strategy based on improved YOLOV5s deep learning algorithm, which can realize the autonomous delivery of the guidewire under image navigation. The development of the vascular model platform experiment proved the feasibility of the scheme.