Modeling, Parameter Identification and Model-Based Control of of Robot Arms
The main topic of this research program is the modelling, parameter estimation and model-based control of CRS A465 robot arm. Modified Denavit-Hartenberg transformation and Newton-Lagrange formulation have been used to find kinematic and dynamic models of the robot. Since the accurate dynamic parameters of robot arms are not included in the manufacturer data, they are estimated by experimental parameter identification. Particular attention is given to the excitation trajectory design as it strongly influences the quality of the estimation results. The torque prediction results indicate the high accuracy of the estimated model. Then an end-effector trajectory tracking is conducted for a challenging trajectory using a model-based inverse dynamic controller. The maximum and mean tracking error was 0.25 and 0.08 mm which shows the high performance of the designed controller [1]. Now the model of the robot is available for other high performance and robust controller designs. One can also evaluate the performance of the controllers in simulation in Matlab/Simulink before testing them in experiments. A user friendly simulink simulator was designed for this purpose for those who do not have much knowledge about robot models but want to design controller for robots.
|