Experimental Study for Controller-Friendly Contact Estimation for Humanoid Robot

Abstract : In this paper, we introduce a practical contact observer which allows detecting a contact on the body surface of a humanoid robot. Our method estimates the error force due to undesired contact. Then, contact detection (yes/no) is performed together with the estimation of the contact force and location. By comparing the real robot state according to sensory data with the desired state computed from the contact-free dynamics equation, we can estimate the external force applied to an unknown link. The link in contact is then detected by tracing the torque error along the robot's kinematic chain. Once the link in contact is determined, we can estimate the point where the contact force was applied by using the 3D geometric mesh model of a robot. The proposed method is validated through simulation and experiments using the humanoid robot HRP-4.
Document type :
Conference papers
Complete list of metadatas

Cited literature [27 references]  Display  Hide  Download

https://hal-lirmm.ccsd.cnrs.fr/lirmm-02304148
Contributor : Kheddar Abderrahmane <>
Submitted on : Thursday, October 3, 2019 - 1:48:26 AM
Last modification on : Saturday, October 5, 2019 - 1:14:59 AM

File

ARSO19_0026_FI.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : lirmm-02304148, version 1

Collections

Citation

Takahiro Ito, Ko Ayusawa, Eiichi Yoshida, Abderrahmane Kheddar. Experimental Study for Controller-Friendly Contact Estimation for Humanoid Robot. ARSO: Advanced Robotics and Its SOcial Impacts, Oct 2019, Beijing, China. ⟨lirmm-02304148⟩

Share

Metrics

Record views

41

Files downloads

80