Joint Stiffness Identification Of 6R Industrial Robot And Experimental Verification

Currently, the industrial robot has been applied to many kinds of industries due to its programmability, adaptivity, flexibility and low cost. And the industrial robot is already applied to machining tasks. However, it has not seen many success stories for such applications. The reason is that the stiffness of industrial robot is much lower than that of a traditional CNC machine. The stiffness for a typical articulated robot is usually less than1N/μm, while a standard CNC machine center often has stiffness greater than 50N/μm.This paper mainly analyzes the stiffness of flexible industrial robots in order to improve robotic machining performance.Firstly in this paper considering KUKA KR16 robot's geometrical structure, we establish the kinematics of the robot by means of D-H method. And then, detailed kinematics inverse solutions are calculated, and the Jacobian matrix is also solved. We also analyze two types of singular configuration of the robot when considering the determinant of the Jacobian matrix equals to zero.Secondly aiming to increase the robot's absolute accuracy, we analyze the error sources which mostly influence the robot's absolute accuracy. And then, the joint stiffness model is established to consider the effect of elastic error in industrial robot, which is conducted by the principle of that the model should be accurate enough for the prediction of robot structure deformation under arbitrary load conditions and it needs to be simple enough for real time implementation. And we propose one observation strategy for joint stiffness identification, which can make sure each unknown parameter of joint stiffness is observable.Then an experimental method is designed to identify the joint stiffness, which is composed by the API laser tracker and the cable pulley system. The external load is exerted on the end point of the robot by way of the cable pulley system and the displacement of the end point is measured via the API laser tracker. And then the joint stiffness is identified through the least squares method. The identification results are proved to be effective due to very small residual errors.At last another experiment is implemented to give deformation compensation of the end point under the external load. The dead weight is exerted on the end point of the robot. Based on the stiffness model identified and the known payload before, the deformation due to the payload is calculated in MATALAB and the joint reference for the robot controller is updated. And we find the compensation results are proved to be effective.