A Real-time 3D Shape Sensing Algorithm Based on Self-Organizing Map

Due to the complex motion structure, it is often difficult to achieve precise control of the soft manipulator. An important part of this is the real-time feedback of the shape of the target. In this paper, a real-time 3D shape sensing algorithm based on stereo vision and Self-Organizing Map (SOM) network without marking points is proposed. The algorithm extracts the edge data of the target for each frame, and uses SOM to cluster the key points at the center line, and then reconstructs the key points in three dimensions to obtain the shape and position information of the target. Experiments is carried out to verify the effectiveness of the algorithm and errors are analyzed.