Vision-Based Shape Mesaurement Of Cable-Driven Flexible Manipulator And Target Position Mathod

Application scenarios such as disaster rescue,circumstances detection,equipment overhaul and maintenance own the characteristics of small working space and many obstacles.It is hard for traditional robots to perform tasks,however the cable-driven flexible manipulator can play an important role because of its flexible movement ability and slender body.Nevertheless,due to large number of transmission links,complicated friction characteristics and super-redundant characteristics,the shape and end accuracy of cable-driven flexible manipulator are low,which restricts the task execution performance.To solve the above problems,this paper studies the vision-based flexible manipulator shape measurement,kinematic parameter calibration and pose measurement method for typical charging target on equipment maintenance circumstances,and develops the corresponding algorithm and conduct experiment for verification.According to the characteristics of cable-driven flexible manipulator,such as lots of moving part,small size and complex spatial configuration,the flexible manipulator shape and end pose synchronization measurement system based on global vision and hand-eye vision fusion is designed to solve the measuring problem of large-scale and high-accuracy.The system consists of two fixed monocular cameras,a hand-eye camera mounted at the end of the flexible manipulator and moving with it,and a task board fixed in the environment.Every moving parts flexible manipulator and the task board is equipped with a two-dimensional code to form a multi-point scattered measurement mark.With the global vision camera identifies the mark on the manipulator,the hand-eye camera identifies the mark on the task board,and then fuses all the mark information,further calculates the joint angle of each flexible arm and reconstructs the space shape of the manipulator.At the same time,the end pose of the flexible manipulator is also obtained,so that the flexible manipulator can sense its state more accurately and help to improve the motion accuracy.A simplified end pose error model is established for the cable-driven flexible manipulator because it consisting of multiple PY(Pitch-Yaw)or YP(Yaw-Pitch)doubledegree-of-freedom joints and its adjacent joints using "PY-YP" connection.A calibration method for kinematic parameters of a flexible manipulator based on vision is proposed.The kinematics modeling of the flexible manipulator is carried out by using the MDH method.The influence of each parameter on the end pose error is analyzed in combination with the joint structure characteristics.The negligible parameters are eliminated and the simplified error model is obtained.Further,serval manipulator calibration shapes covering a large working space are designed,and use the vision measurement system to obtain the shape and end pose of each arm type,and this is used for the error model to calibrate the kinematic parameters.The calibration results of the laser tracker are compared to verify the feasibility of the method.Taking the maintenance of aviation equipment as an application case,the pose measurement method based on natural feature detection of the actual target is proposed.After extracting multiple corner points and circular features,the PnP algorithm is used to realize the pose measurement at long distance and close distance.The target is imaged by hand-eye vision.The target is detected by using muti-template that considering both near and far distances and different perspectives to obtain an effective target area.Corner and center features are extracted in the area,in further the corner points and center feature are respectively combined with the PnP algorithm to obtain the target pose according to the distance.Finally,the vision measurement experimental system of flexible manipulator is built.Synchronized measurement of the shape and end pose of flexible manipulator,calibration of the kinematic parameters of the flexible manipulator,target pose measurement and docking experiments are carried out to verify the above methods.