Research On Environment Parameter Estimation And Compliant Control Of Force Feedback Teleoperation

Using robot instead of human to work in the condition that is dangerous or difficult to reach can both save the cost and avoid human injury. However, within the limitation of sensor device, control system and machine intelligence, it is still difficult to develop the autonomous robot that can perform complex tasks in unstructured workspace.Consequently, the force feedback teleoperation is still the principal tool for executing robot tasks within special area in the foreseeable future.As one of the three main approach of force feedback teleoperation,the model-mediated teleoperation has been paid more and more attention in recent years because of its robustness to large time delay, remarkable performance on both stability and transparency. In this thesis, the environment parameter estimation technique and compliance control in slave side were analysis and investigated. The main work of this thesis is as follows.Firstly, different modeling methods and parameter estimation algorithms are analyzed and tested. Cross-validation is utilized in the comparison of different estimation algorithms to measure model precision quantitatively.Secondly, a new model updating algorithm is proposed on the basis of analyzing and comparing the existing methods. The proposed model updating algorithm defines a sudden force and try to control the force rendering under the constraint of it. Although the aim of both proposed algorithm and the exist methods is to make the force rendering smooth even though the parameters in prediction model is changing sharply, the proposed approach control the force directly instead of the model parameters. Significant improvement on both efficiency and stability is showed in experiment.Thirdly, an adaptive impedance controller is proposed, which can be adopted in the salve side of model-mediated teleoperation. The risk of overlarge contact force or penetration of hybrid force/position control and relative tracking method was pointed. The new impedance controller was developed for this problem. By using desire force and desire position as the reference input of the controller, the slave is able to tracking the desire position while the prediction model in master is accurate. The adaptive law of impedance parameter were derived based on the ratios of force and penetration on slave side to that on master side. The slave controller is able to track the command and maintain stability on slave side in the presence of large error between prediction model and the real environment.Finally, the approaches proposed in this thesis are verified experimentally in a model-mediated teleoperation system. The experiment further prove the effectiveness and robustness of the proposed methods.