Random vector functional link network with L21 norm regularization for robot visual servo control with feature constraint

Uncalibrated visual servoing control still encounters some challenges, such as calculating the interaction matrix with less cost and keeping the current image features within a camera’s field of view (FOV) in a noisy system environment. To solve these problems, we propose a new control method that u...

Full description

Saved in:
Bibliographic Details
Published in:Journal of mechanical science and technology 2022, 36(9), , pp.4747-4759
Main Authors: Zhou, Zhiyu, Guo, Jiusen, Wang, Yaming, Zhu, Zefei
Format: Article
Language:English
Subjects:
Algorithms
Cameras
Control
Control methods
Control theory
Dynamical Systems
Engineering
Field of view
Industrial and Production Engineering
Mathematical analysis
Mechanical Engineering
Original Article
Regularization
Robot control
Servocontrol
Vibration
Visual control
기계공학
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Uncalibrated visual servoing control still encounters some challenges, such as calculating the interaction matrix with less cost and keeping the current image features within a camera’s field of view (FOV) in a noisy system environment. To solve these problems, we propose a new control method that uses a random vector functional link network with L21 norm regularization to calculate the interaction matrix and further estimate it with a robust information filter (RIF). L21 norm regularization can deal with the global sparsity of input weights and reduce the inherent complexity of a model. The RIF limits noise variance within a certain range to reduce the influence of uncertain noise on the servoing task. We also design a method that reacts to the control law in accordance with the coordinates of image features. It can adjust running speed in real time and keep image features within a camera’s FOV. We apply this method to a six-degrees-of-freedom eye-in-hand manipulator, and several simulations are performed. Simulation results show that the proposed algorithm performs well in the task and achieves good performance in terms of noise resistance. Image features barely escape from the camera’s FOV through the proposed constraint method.
ISSN:1738-494X
1976-3824
DOI:10.1007/s12206-022-0834-y