The Visual Navigation Method Of Double Hemisphere Capsule Robot

The magnetically driven capsule endoscope is a research hotspot in the field of international micro-robots.Our group had proposed a kind of double hemispherical capsule robot driven by spatial universal rotating magnetic field(SURMF)and its turning drive control method.The robot is suitable for operation in ample environment such as the colon which solves the key technology of separating the two modes of dynamic attitude adjustment and rolling locomotion,realising the hovering of the capsule.However,after the double hemispherical capsule enters the human body,the position and posture information of the capsule is unknown,so even with the aid of real-time video assistance,the navigation orientation cannot be determined.At present,the turning navigation of the capsule can only adjust the capsule posture by means of multiple attempts until the alignment of the bowel direction is approximated by the image which has poor operability,low precision and efficiency and cannot give direct navigation direction to the next turning motion of the capsule.To realize the steering locomotion of dual hemisphere capsule robot in curved intestinal tract,a visual navigation method of double hemisphere capsule robot is proposed in this paper.First,determine the direction of the axis of the capsule based on the coaxial follow-up characteristics of the capsule axis and the spatial universal rotating magnetic vector during dynamic posture adjustment process.Two methods are proposed to obtain the rotation angle of the capsule robot around the capsule axis,so as to determine the attitude information of the capsule robot.Second,using the regional uniformity of the three-axis Helmholtz rotating magnetic field,combined with the attitude information of the capsule robot and the centroid of the dark region on the wireless transmission image,the coordinate system relationship is established and the coordinate transformation is performed to obtain the final navigation direction,thus the visual navigation of the capsule in the curved intestine tract can be achieved by further calculating the direction of the the spatial universal rotating magnetic vector for rolling locomotion.In order to accurately extract the centroid of the dark region of the image,a dark region centroid recognition calculation method based on the improved BAT algorithm and the improved two-dimensional OTSU algorithm is proposed.The improved two-dimensional OTSU algorithm can improve the problem that the traditional OTSU and two-dimensional OTSU algorithms are susceptible to folds and gray-scale unevenness of the intestinal wall,which the problem of the slow speed of the algorithm coming along with can be solved to some extent by the improved BAT algorithm.Finally,in order to verify the feasibility of the visual navigation method of the double hemisphere capsule robot,a capsule simulation experiment device was designed and tested in the isolated pig intestine.The experiment proves that there is a small error between the navigation direction calculated by the method and the theoretical value,but the navigation method can be used for the navigation of the capsule robot due to the fault tolerance of the double hemispherical capsule robot rolling in the intestine.The method satisfies the requirements of small impact on the overall volume,strong anti-interference ability,high safety to the human body,simple processing and assembly process,low requirements for installation precision,simple operation,strong real-time performance and meeting navigation accuracy,which lays a foundation for all-over inspection and medical operation inside the GI tract of human body.