ViO-Com: Feed-Forward Compensation Using Vision-Based Optimization for High-Precision Surgical Manipulation

Cited 5 time in webofscience Cited 0 time in scopus
  • Hit : 289
  • Download : 0
Tendon-sheath mechanisms offer a means for a flexible surgical robot to be operated efficiently in restricted environments, for example, the long and narrow paths inside human organs. However, nonlinear hysteresis interferes with the precise motion control of a flexible robot. Generally, the characteristics of hysteresis can change due to changes in cable-related effects and robot shapes, which increase the difficulty associated with achieving precise control of surgical robots. Although several methods have been proposed to solve this problem, most of these methods offer limited performance and are unstable in realistic situations. In this letter, we present a vision-based optimized feedforward compensation (ViO-Com) scheme for a cable-driven surgical robot that reduces hysteresis as a practical approach, regardless of any unmeasured hysteresis while maintaining stable motion. A CycleGAN and a Siamese CNN were used to estimate the actual joint angle of a surgical manipulator, and the Bouc-Wen model with optimized parameters was used for feed-forward compensation. The results obtained using a real surgical robot platform K-FLEX suggest that the performance of ViO-Com is superior to that of vision-based feedback compensation under various hysteresis condition, and its accuracy is improved by 67%.
Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
Issue Date
2022-01
Language
English
Article Type
Article
Citation

IEEE ROBOTICS AND AUTOMATION LETTERS, v.7, no.1, pp.263 - 270

ISSN
2377-3766
DOI
10.1109/LRA.2021.3123375
URI
http://hdl.handle.net/10203/289653
Appears in Collection
ME-Journal Papers(저널논문)
Files in This Item
There are no files associated with this item.
This item is cited by other documents in WoS
⊙ Detail Information in WoSⓡ Click to see webofscience_button
⊙ Cited 5 items in WoS Click to see citing articles in records_button

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0