Dynamic identification of the Novint Falcon Haptic device

Abstract

The main concern of this paper consists in dynamic identification of the so-called “Novint Falcon” parallel Haptic device. In this regard, a precise CAD model of the under study robot is designed and dynamic properties of its constituting parts are assigned to the model. Using a white box identification procedure, the unknown parameters in closed-form dynamic model of the robot is identified in the so-called “SimMechanics” simulation environment of Matlab. To this end, by persistently exciting robot input motions, i.e., sets of sine wave functions with different frequencies and amplitudes, the required torques for actuators is calculated by SimMechanics. Benefiting from Least Square (LS) method for given model structure and collected actuators motion and torque data, dynamic parameters are determined. For the sake of verification, the motion and torque data obtained form identified dynamic model is compared with SimMechanics results for a specified trajectory and the Root Mean Square Error for actuators is about 0.3 N.m. Therefore, it can be inferred that identified model is acceptable in the terms of accuracy.