Robot Arm End Pose Measurement Based On Multi-sensor Information Fusion

For fulfilling the task,the service robots need to control the end effector to reach the specified position,which requires real-time measurement of the end pose.In order to improve the pose measurement accuracy of the serial robot arm with lower cost sensors,this paper studies the technique of end pose measurement of the seven-degree-of-freedom rehabilitation robot arm with multi-sensor information fusion.After analyzing the advantages and disadvantages of the methods with single category sensor,the inertial sensor and the visual sensor are combined to design a real-time,fast and accurate robot arm end pose measurement method.The main research contents are as follows:Firstly,this paper analyzes the development and background significance of the topic and the popular pose measurement methods.The feasibility and main scheme of the research are determined based on the development status at home and abroad and the application platform.Secondly,two kinds of pose measurement methods based on inertial sensors are studied,which include the compensation of the sensor installation error by using a fitting circle to find the rotation matrix between the sensor coordinate and the robot arm coordinate.Then the end pose is obtained by utilizing the D-H model with joint angles that is measured by the IMU sensors.The other method to obtain the end pose is measuring by the IMU fixed at the end.Thirdly,the pose measurement based on monocular vision is studied.A hand-eye calibration method based on circular path is proposed to determine the the transformation matrixs of target coordinate and camera coordinate with robot arm coordinate with orthogonal joints for an eye-to-hand system.Then using the‘Harris'algorithm to extract the point features of the target,and the pose is solved based on PNP method.Finally,the multiple model fusion algorithm based on Kalman filter and H~?filter is studied.The Kalman algorithm combines two inertial pose measurements and the H~?filter algorithm combines data measured by IMU sensors and camera.And it is verified by experiments that the fusion method can effectively solve the problem of drift error and also improve the problem of small measurement range.It realizes the complementarity of inertial sensors and vision sensors,and improves the accuracy of the robot arm end pose measurement.