Shape Detection Method And Experimental Research Of Continuum Robot Based On Stereo Vision

The continuum robot has shown great application prospects in the development of minimally invasive surgery.Its inherent structural flexibility makes it possible to reach the lesion location along the tortuous path in a confined space,and it can perform complex and delicate surgical procedures inside the patient's body.However,the inherent flexibility of the continuum robot and the uncertainty caused by the contact between the instrument of continuum robot and human tissues during the operation make its shape unknown,which makes it difficult to accurately describe the motion and may cause the risk of tissue damage.Therefore,accurate real-time three-dimensional shape detection for continuum robots is necessary.Based on the stereo vision technology and the contour features of continuum robots commonly used in minimally invasive surgery,this paper realizes real-time three-dimensional shape detection of continuum robot based on the markers and the self-organizing map(SOM)neural network algorithm respectively,and designs the offline and real-time detection software platform accordingly.The main research contents of this paper are as follows:Firstly,the continuum robot shape detection platform is built,and the calibration platform system calibration is completed.The system calibration includes stereo camera calibration and the calculation of the transformation matrix between the camera coordinate system and the continuum robot coordinate system.The camera model and system calibration method are introduced.The system calibration results are obtained by means of the MATLAB calibration toolbox,and the accuracy of the obtained transformation matrix between the camera and the continuum robot coordinate system is verified.Secondly,real-time three-dimensional shape detection of continuum robot based on the markers is realized.On the basis of marker extraction and match,two threedimensional reconstruction methods based on triangulation and optimization are realized respectively.In addition,based on the SOM algorithm,a real-time shape detection method without setting markers on the robot is realized,which improves the the algorithm efficiency compared with that of using the traditional SOM method.Based on the Visual Studio 2013 development environment,the interface programming is performed with the MFC framework.The OpenCV and OpenGL development tools are applied to design the corresponding offline and real-time detection software for the proposed marker-based and SOM-based shape detection algorithm.In this process the human-computer interaction and user-friendliness are considered.Finally,the experimental research on the algorithm proposed in this paper is carried out to verify the detection accuracy,error stability and program real-time performance of the robot shape detection algorithm based on the markers and SOM method.The experimental results show that both the proposed algorithms can achieve real-time three-dimensional shape detection of continuum robot with high accuracy.