Japan Advanced Institute of Science and Technology

Japan Advanced Institute of Science and Technology

JAIST Repository

https://dspace.jaist.ac.jp/

Title

Author(s)

Citation

Issue Date

Type

Text version

URL

Rights

Description

with Integral Action

Yoshinobu Nakao

Scho ol of Information Science,

Japan Advanced Institute of Science and Technology

February 13, 1997

Keywords: robust visual servo,integralaction, industrial robotmanipulator.

In this thesis, we deal with visual servo control law with integral action in 2DOF

(TwoDegrees OfFreedom)planarmanipulator. Wepresentthe Lyapnov stabilityanaly-

sisincludingthemanipulatordynamicswithconstanttorquedisturbancesintotheoverall

visual closed loopsystem. Further,weconstruct an experimental system usinganindus-

trial manipulator. Finally, experimental results are shown to illustrate the controller

performance.

Recently, manipulator control schemes that utilize the visual information in the real

time feedback, called visual servo, have attracted much attention. The manipulator is

included into the visual closed loop system. Accordingly, it is important to perform the

stability analysis ofthe visual system with the nonlinearmanipulator dynamics.

Forthe jointspace control,the recentapproacheshavebeen concernedwith theSP-D

(Saturated Prop ortional and Dierential) control. The SP-D control method has some

practicalcharacteristics, forexample,attenuationofundesired hightorque andreduction

of the undesired eect of friction. However,when there exist the constanttorque distur-

bances,thestabilityanalysiscannotbecarriedoutusingtheSP-Dcontrolmethod. Thus,

an integral action was added to the previous SP-D controller in order to deal with the

constant disturbances. Furthermore, it was intro duced that the integral action reduces

steady state errors. The SP-D control has been successfully applied to the visual servo

controlproblem. However,thevisualSP-Dmetho ddoesnotprovidethestabilityanalysis

in the case of the existence of the constant torque disturbances.

Therefore, in our thesis, the image based visual servo control for the 2DOF planar

manipulatorwiththeeye-in-handcongurationistreatedbyinvokingtheLyapunovbased

method. Thecontrolmetho disthe SP-Dvisual servocontrollawwithintegralactionand

thegravitycompensation. Theproposed controlmethodisimprovedtob erobustagainst

Copyright c

1998byYoshinobuNakao

issimilartothe jointspace SP-Dcontrollercase. Furthermore,theproposed SP-Dvisual

servocontrol law with integral actionis implemented onthe industrial manipulator,and

we conrmthe previous controller's availability.

This paper is organized as follows. In section 2, an experimental setup and some

physical parameters of the systems are described. The relations among the set of the

joint angle and the position and the orientation of the camera become clear. Then the

Jacobian of the manipulator is derived. It show the relation between the velo city in

angularspaceand thevelo city inCartesian space. Furthermore,acalibrationmethodfor

the physicalparametersof the manipulatoris described.

Section 3formulatesthe visual servo control problem. Weformulatesome mo dels. It

has the pin hole camera model, the manipulator kinematicsmodel and the manipulator

dynamics mo del. Furthermore,westate the visual servo controlproblem.

In section 4, rstly, we show that, the stability analysis can not be carried out using

the SP-Dvisual servocontrol metho dwhen thereexistthe constanttorque disturbances.

Secondly,weproposetheSP-Dvisualservocontrollawwithintegralactionandthegravity

compensationfor solution of the problem. Finally, we prove the asymptotic stability of

the proposed SP-Dvisual servocontrollawwithintegralaction. The pro ofofthe control

methodisappliedtothemethodsofthejointspaceSP-Dcontrollawwithintegralaction.

Moreover,the visual servocontrol methodisstated to berobust against the focal length

of lens of the camera, the distance b etween the object and the camera and the constant

torque disturbances.

Usefulness of the proposed visual servo control law is veried by the experiments in

Section 5. First, we compare the proposed visual servo control law and the SP-D visual

servocontrol lawwith the gravity compensation. Atthat time, positiongain error inthe

proposed visual servo control law is decided smaller than the SP-D visual servo control

law with the gravity compensation. Second, weexperiment with in case of the constant

torque disturbances. Furthermore, we show the experimental results by some graph. As

well, weshowthe considerationof the experimental results.

Finally,briefconcludingremarksareoeredinsection6. Weshowedthat,thestability

analysis cannot b e carried out using SP-D visual servocontrol method when there exist

the constant torque disturbances. We prop osed the SP-D visual servo control law with

integral action and gravity compensation for solution of the problem. We described the

architecture of the target system and some physical parameters of the systems. The

visual servo control method was improved to be robust against the fo cal length of lens

of the camera, the distancebetween the object and the camera and the constant torque

disturbances. The proposed visual servo control law was implemented on the industrial

manipulator. And we conrmed the previous controller'savailability.