Research On Force-sensorless Direct Teaching Of Robot

The industrial robot is developing towards light weight, and the human-robot cooperation(HRC) ability is improving, there appears new requirements for the robot teaching techniques accordingly. Compared with the traditional teaching technique by teaching pendant, the direct teaching technology has the advantages of high efficiency, simple operation, low requirements for the operator, thus has become the focus of the development of robot teaching technology. Compared with direct teaching with force sensor, force-sensorless direct teaching has obvious advantages because it can work without the expensive force sensor, yet there are still lots of technical problems to be solved before it can serve us. Thus, this thesis focuses on the research on the force-sensorless direct teaching technology, a scheme based on torque compensation, especially dynamic compensation, is proposed to achieve force-sensorless direct teaching for the industrial robots.In this thesis, based on the analysis of each item in the robot dynamic equation, and in consideration of the characteristics of different robots, two different torque compensation schemes are proposed. Firstly, the scheme based on partial dynamic torque compensation, namely only compensating gravity torque and friction torque during the teaching process, is proposed for the robot which has light weight and small inertial. For the parallel robot, which has light weight and small inertial, the nominal dynamic model is used. We take advantage of the existing robot dynamic model and the measured values of the robot parameters to obtain the nominal model of the robot. Based on nominal model of the robot, we apply torque control by real-time gravity compensation, and apply the scheme on the parallel Delta robot. Experiment results show that the scheme can obtain good compensation effect,the force needed for the operator to guide the robot manipulator moving widely in the whole working space is 5~10N on average, with the maximum value less than 20 N. For the serial SCARA robot, which also has light weight and small inertial, the identification method is used. The modeling of joint gravity torque and joint friction torque is discussed, and the identification experiments are done, after which we achieve the direct teaching of the SCARA robot by joint gravity torque and joint friction torque compensation. Considering the need of direct teaching with load for the serial 6R light-weight robot in industrial application, a new method of online identification of the first moment of mass is proposed to obtain the gravity torque of the robot. Experiment results show that this new method can achieve good online identification for the first three joints, however, not so good for the last three joints. It preliminary verifies the correctness of the theory for the method and the feasibility. Secondly, the scheme based on integral robot dynamic torque compensation is proposed for the 6R robot which has relatively large weight. Robot dynamic model identification is studied, after the procedure of dynamic modeling, reforming the dynamic equation to linear expression, optimizing the excitation tra jectory, parameter estimation and validation, we obtain the accurate dynamic model of the robot. And finally, we apply the dynamic compensation scheme on our open control 6R robot platform to achieve the direct teaching. Experiment results show that it has achieved good quality and user experience. We also develop and improve the software part of the direct teaching system.