Napredna pretraga

Pregled bibliografske jedinice broj: 305978

Modified Broyden method for noise visual servoing


Bonković, Mirjana; Hace, Aleš; Cecić, Mojmil
Modified Broyden method for noise visual servoing // Proceedings of 6th EUROSIM Congress on Modelling and Simulation / Borut Zupančič, Rihard Karba, Sašo Blažič (ur.).
Ljubljana: ARGESIM - ARGE Simulation News, Div. Mathematical Modelling and Simulation, Vienna University of Technology, 2007. (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)


Naslov
Modified Broyden method for noise visual servoing

Autori
Bonković, Mirjana ; Hace, Aleš ; Cecić, Mojmil

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of 6th EUROSIM Congress on Modelling and Simulation / Borut Zupančič, Rihard Karba, Sašo Blažič - Ljubljana : ARGESIM - ARGE Simulation News, Div. Mathematical Modelling and Simulation, Vienna University of Technology, 2007

ISBN
978-3-901608-32-2

Skup
6th EUROSIM Congress on Modelling and Simulation

Mjesto i datum
Ljubljana, Slovenija, 9-13.09.2007

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Uncalibrated visual servoing; Jacobian estimation; Broyden matrix

Sažetak
Uncalibrated, model free, robot visual servoing has been widely applicable in robot vision due to minimal requirements related to calibration and robot kinematic’ s parameters. The numerical quasy-Newton methods offer a theoretical background for problem solving, which has been proven hard as the real system has been influenced with the noise. Consequently, additional attention has to be paid which assured stability and the robustness of the proposed method. In this paper we have introduced a new, modified Broyden method for nonlinear optimization problem solving. Modified Broyden method has been achieved applying the variable parameter, which values depend on the Broyden matrix convergence condition. In this paper the standard nonlinear optimization technique (Broyden methods for nonlinear equation solving) has been adopted according to Broyden matrix convergence condition and applied for visual servoing control problem which, due to the noise regularly present in the real systems, could be hardly control by the pure Broyden itself. The developed algorithm is verified by simulations for uncalibrated vision-guided robotic control and compared with four methods from literature, which have usually been used for the similar purpose. Modified Broyden method shows performance improvement over previous methods and is more robust in the presence of noise.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo



POVEZANOST RADA


Projekt / tema
023-0232005-2003 - AgISEco - Agentski orijentirani inteligentni sustavi nadzora i zaštite okoliša (Darko Stipaničev, )

Ustanove
Fakultet elektrotehnike, strojarstva i brodogradnje, Split