Research On Motion Control Of Cable-driven Snake Manipulator Based On Vision

The snake manipulator is a new type of biomimetic continuous robot.Compared with the traditional tandem type manipulator,it has a more flexible motion mode,and the drive element and the moving part are spatially separated from each other,which has higher working efficiency.Therefore,it has broad application prospects in space target acquisition,strong radiation environment detection,aviation engine maintenance,and disaster rescue.The snake robot arm relies on the rope to drive the bionic joint movement,which has higher structural rigidity than the fully flexible manipulator,but at the same time maintains a high degree of freedom of motion.Based on the biological spine structure,this paper refines its key structure,develops a new type of snake robotic arm robot,and conducts a comprehensive kinematics analysis.On this basis,it adds visual joint angle recognition and binocular.The target tracking link constitutes a visual servo system.According to the structural characteristics of the biological spine,the robot designed in this paper adopts the rope driving method,which is formed by connecting five bionic joints under the joint of the universal joint for 10 degrees of freedom and 15 driving ropes.Based on the structural design,the D-H parameter is used to express the pose of the robot.From the single joint to the whole arm segment,the forward and inverse kinematics are analyzed.Considering the driving characteristics of the snake manipulator,a method for determining the relationship between the rope speed and joint velocity based on the joint force rotation is proposed.On the basis of the above work,the control framework is proposed,and the rationality of the above analysis is verified by simulation means.In order to solve the zero point determination problem of the snake arm robot,a vision-based method for measuring the joint angle of the mechanical arm is proposed.Based on the basic theory of machine vision such as camera model,calibration method and hand-eye calibration,the position of the camera is solved by positioning the Aruco identification code to identify the joint angle.In the process of zero return,the error between the real-time joint angle and the zero position is obtained through the PD link to obtain the desired joint angular velocity,and then the rope speed is obtained through the joint angular velocity-rope velocity mapping relationship,so that the robot arm has the autonomous zero-return capability under visual assistance.The accuracy of the initial pose of the manipulator is guaranteed,which makes it have higher motion accuracy.The tracking ability of space targets is extended to some extent.The application of the snake-shaped manipulator is extended to some extent.Based on the Camshift algorithm,a binocular target tracking algorithm is proposed.The binocular camera fixed at the end is used to determine the spatial target position.The end pose is measured to obtain the end pose deviation.Finally,the velocity gradient optimization method is used to obtain the joint angular velocity expectation,and then the robot arm is controlled to track the space target.On the basis of the above work,the physical prototype is developed,and the experimental verification and performance analysis of the algorithm for the large-scale movement of the snake-driven snake-shaped arm,the open-end end trajectory planning,the visual zero return,and the end target tracking are completed.The experimental results show that the motion analysis of the snake manipulator is correct,the prototype design is reasonable,the movement is flexible and stable,and it has the ability of autonomous zero return and end target tracking.