A Stepwise Identification Method for Dynamic Parameters of 2-DOF Serial Wrist

In order to establish an accurate dynamic model for a 2-DOF serial wrist, a stepwise identification method for dynamic parameters is proposed on the basis of characteristics that a periodic input results in the periodic variations but different phases of the inertia and gravity term and friction force. This method takes the typical periodic motion commands as inputs, such as sine/cosine function. Based on the integral-elimination method, a reasonable time interval for the numerical integration needs to be designed to identify friction parameters. Subsequently, the friction torque is extracted from the sampled torque. In this paper, the genetic algorithm is used to identify the inertia parameters with the rigid-body dynamics model of the 2-DOF serial wrist. Eventually, the 2-DOF A/C serial wrist installed on a hybrid robot is used in the verification experiments. Experimental results show that the driving torque predicted by the identified dynamic model is consistent with the measured torque and the validity of the stepwise identification method for dynamic parameters proposed is verified.