Optimal exciting motion for fast robot identification. Application to contact painting tasks with estimated external forces - LIRMM - Laboratoire d’Informatique, de Robotique et de Microélectronique de Montpellier Access content directly
Journal Articles Robotics and Autonomous Systems Year : 2019

Optimal exciting motion for fast robot identification. Application to contact painting tasks with estimated external forces

Abstract

Accurate geometric and inertial parameter estimates of a modern manipulator are of crucial importance to obtain good performances during a contact task or for obtaining more and more required realistic simulations. CAD data are often provided by the manufacturer, but these are inaccurate and do not take into account eventual end-effector modifications. Fortunately, they can be identified. However, in real industrial applications, dynamic identification is rarely performed because it supposedly requires a cumbersome and long procedure. There is a need of a practical but accurate method to identify dynamics parameters. Thus, this paper proposes a practical framework to identify a Kuka LWR robot in less than 10 s. An experimental comparison between several cost functions showed that is the best trade-off for getting a good parameters accuracy within a minimal time. The procedure identifies very accurately the inertial parameters of the robot and of its end-effector and recognizes its geometric parameters from a look-up table. When using identified parameters, joint torques were estimated with an RMS difference lower than 1 N m when compared to measured ones. The identified model was then used to generate a contact painting trajectory. During this contact task, the external forces were estimated and controlled without the use of a force sensor. Experimentation showed that the external forces could be identified with an RMS difference lower than 3 N.
Fichier principal
Vignette du fichier
S0921889017307091.pdf (1.71 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

lirmm-02011328 , version 1 (21-10-2021)

Licence

Attribution - NonCommercial

Identifiers

Cite

Takuma Katsumata, Benjamin Navarro, Vincent V. Bonnet, Philippe Fraisse, André Crosnier, et al.. Optimal exciting motion for fast robot identification. Application to contact painting tasks with estimated external forces. Robotics and Autonomous Systems, 2019, 113, pp.149-159. ⟨10.1016/j.robot.2018.11.021⟩. ⟨lirmm-02011328⟩
151 View
81 Download

Altmetric

Share

Gmail Facebook X LinkedIn More