An Excitation Trajectory for Dynamic Parameter Identification of Robot Using Logistic Function and Depth-First Search

Abstract

Abstract The dynamic parameter identification is playing an important role in robot dynamic controller design. However, many excitation trajectories for parameter identification cannot cover enough motion states of the robot joints, which weakens the generalization of the dynamic model. In this paper, an excitation trajectory covering multi-axis motion states based on depth-first search (DFS) is proposed and spliced from logistic functions, in which the angular velocity and angular acceleration can be directly obtained from joint angle by the analytical formula. Taking the condition number of observation matrix as an optimization index, the excitation trajectory is optimized based on genetic algorithm. The strategy of iterative identification is adopted, and in each iteration the dynamic parameters are identified according to the last position and torque sequences and updated to the controller until the actual trajectory approaches the desired trajectory. Experimental results show that after two iterations, the tracking accuracy is greatly improved, and the effectiveness of the proposed approach is verified.