Experimental identification of the dynamics model for cartesian robot

This paper proposes an experimental-based approach to estimate the kinematic parameters and develop a cascade controller of a Cartesian robot. The aim is to satisfy the position accuracy in the trajectory execution, guaranteeing a high dynamics. The proposed procedure consists of a set of experimental tests executed on a reference trajectory, varying the velocity and acceleration in a specific range. The model is based on the Least-Square-Estimation and the Genetic Algorithms. Once the kinematic parameters have been calculated and evaluated, the controller has been built using an iterative procedure to estimate the PID gains of the position, velocity and current loops. Finally, the overall system has been validated through a set of reference trajectories, comparing the observed results with the predicted values. The RMSE index of torque has shown a congruence between the obtained results with a maximum value lower than 8.0 Nm.