Use this URL to cite or link to this record in EThOS:
Title: Computationally intelligent visual servoing
Author: Siradjuddin, Indrazno
ISNI:       0000 0004 5353 3986
Awarding Body: University of Ulster
Current Institution: Ulster University
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
In visual servoing methods, the pseudo-inverse of the coupled robot-image Jacobian plays important role which maps the error in image space to the joint velocity. This map is highly non-linear and computationally complex. The complexity increases in order of O(N2) where N is the number of robot-manipulator joint. Another problem with classical visual servoing methods or model based approaches is that the system instability may occur when a discrepancy between reference model and the real system exists. This thesis proposed visual servoing methods based on computationally intelligent techniques addressing problem in the model based visual servoing approaches. A model free image based visual servoing approach using a modified Broyden's method is proposed in this thesis, The robot-image Jacobian is approximated iteratively, in each instant, during servoing towards the desired position. In addition, a computationally simple robot-image Jacobian approximation is applied together with a weighted norm method for joint limit avoidance. A computationally efficient approach for image based visual servoing has also been developed using the Takagi-Sugeno (T-S) fuzzy model. In a distributed fashion, the proposed approach maps directly the error in the visual space to the joint velocity. The method can tune the controller's parameters online. The Lyapunov method is used to derive the online adaptation algorithm. Finally a position based visual servoing approach using the Kalman filter is proposed and evaluated for a visual tracking system. This proposed method uses both the position error and its derivative as the input variables to the controller. A Kinect camera is used to provide 2D and 3D information of the object. Using a Kinect camera, the controller design effort can be simplified, while the approach reduces the computational complexity of the position based visual servoing algorithm, since the complex 3D estimation algorithm is not required. The experiments reported in this thesis work use a 7 DOF PowerCube robot manipulator. A stability analysis for the proposed algorithms using the Lyapunov method is provided.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available