Identifying Dynamic Parameters With a Novel Software Design for the M-DOF Collaborative Robot

Abstract

The primary goal of this project was to develop a general identification method via software that can be applied to collaborative robots. To achieve this, the collaborative ultralight robots Kinova Gen2 and Kuka LWR4+ with seven degrees of freedom (M-DOF) were used. Specifically, the “recursive Newton-Euler” formulation was used to provide a set of parameters that could describe the body structure and to create a general symbolic representation for collaborative robots. For parameter estimation, the least squares method was used. In addition, trajectories generated with random numbers typically do not produce consistent results; thus, verified trajectories were used. To verify trajectories, real robots were simulated with V-Rep before being executed. When untested trajectories are first tested on robots, undesirable results may occur. This method was convenient for parameter estimation and robot health; saves time; and increases the consistency of results. Algorithms were coded in MATLAB and ROS packages via Python. MATLAB, ROS, and V-Rep worked together in the Ubuntu operating system. The identification methods were modeled, implemented, tested, and validated successfully, and the results for both robots are reported in this article.