Aiming at the problems of low autonomy and inability to cover the lesion area in the trajectory planning of current ultrasonic robots

an autonomous trajectory planning method based on visual point cloud information is proposed with cardiac ultrasound examination as an example.Firstly

a depth camera is used to locate the patient’s scanning area and extract the point cloud information by using the human keypoint recognition method.Then

an isometric 3D point cloud search algorithm is used to extract waypoints from the preprocessed point cloud data to achieve uniform coverage of the scanning area

and the normal vectors of the waypoints are obtained.Finally

the complete pose description of the waypoints is obtained by deriving the coordinate transformation relationship of the system and modeling the local coordinate system of each waypoint.Trajectory planning is completed using cubic spline curves and S-shaped velocity curves.Experimental results show that this method can realize end-to-end fully autonomous trajectory planning

and reduce the planning time for robotic ultrasound scanning to 16 s

which effectively improves the scanning efficiency.