Learning of Contact Motion Using a Neural Network and Its Application for Force Control  

Oleh:

Nagata, Fusaomi ; Watanabe, Keigo

Jenis: Article from Article
Dalam koleksi: Final Program and Book of Abstracts: The 4th Asian Control Conference, September 25-27, 2002 (Sep. 2002), page 420-424.
Topik: Neural Network; Contact Motion; Force Control; PUMA560; Dynamic Model
Fulltext: AC021597.PDF (148.8KB)

Isi artikel

In this paper, a force control method using a neural network is proposed for multi-degrees-of-freedom (DOF) manipulators. First, a desirable relation between the contact force to workpiece and the approach velocity of the manipulator is created using a conventional PI force control. Next, such a relation is mapped into the neural network. The learned neural network yields the velocity command for the force feedforward control according to the force error and its rate. The proposed force control using the neural network has a promising flexibility to both environmental stiffness and approaching velocity. The effectiveness of the proposed method is examined by some simulations using a dynamic model of PUMA560.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali