Adaptive Visual Servoing With an Uncalibrated Camera Using Projection Homography and Q-Learning
IEEE Access · 2026 · 14 · 121303-121313
DOI: 10.1109/access.2026.3721478 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Meng Kang und 2 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
In this article, a novel image-based visual servoing (IBVS) method based on projection homography and Q-learning is proposed to solve the problem of slow convergence when there are numerous feature points in the uncalibrated visual servoing system. First, the projection-homography-based visual servoing (PHBVS) is designed to solve the problem that the dimension of the interaction matrix increases as the number of feature points grows. Second, the Q-learning method is applied to determine the gain to improve convergence speed without prior experience. Compared with other methods, PHBVS has a fixed interaction matrix that can reduce the computational load when many feature points exist. Furthermore, Q-learning enables self-learning without prior experience and achieves fast convergence. The effectiveness of the proposed method is validated by simulations on a 6-DOF manipulator model with an eye-in-hand configuration.
Zitieren
APA
Kang, M., Cao, J., & Wei, P. (2026). Adaptive Visual Servoing With an Uncalibrated Camera Using Projection Homography and Q-Learning. IEEE Access, 14, 121303-121313. https://doi.org/10.1109/access.2026.3721478
Vancouver
Kang M, Cao J, Wei P. Adaptive Visual Servoing With an Uncalibrated Camera Using Projection Homography and Q-Learning. IEEE Access. 2026;14:121303-121313. doi:10.1109/access.2026.3721478
BibTeX
@article{Kang2026,
author = {Kang, Meng and Cao, Jiangwei and Wei, Peng},
title = {Adaptive Visual Servoing With an Uncalibrated Camera Using Projection Homography and Q-Learning},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {121303--121313},
doi = {10.1109/access.2026.3721478}
}
EUVIMED