A common mistake in the concept of image search is that the technology is based on detecting and processing the information in the image itself. Searching for an images works like this; the meta data of the image is indexed and stored in a large database or repository and when a search query(keywords) is entered the image search engine accesses the index, and queries are matched with the stored information. The results are presented in no particular order of relevancy. The effectiveness of an image search engine depends on the relevance of the results it returns, and the clustering algorithm plays a big role. This paper compares the working of two clustering algorithms K-Means algorithm and Fuzzy C Means algorithm. When using Fuzzy C Means algorithm, one image can appear in more than one cluster unlike K-Means which is hard based grouping. Through the results it can be clearly seen that using Fuzzy C Means brings about a level of flexibility and proper clustering. The user can access images from an image search engine, picture library, trained data sets, etc. Hence there is a necessity of providing the user more accurate collection of images which can be done through Fuzzy C Means clustering.

This paper uses Fuzzy K-Means clustering algorithm to access images from a collection of images. When using this algorithm one image can appear in more than one clusters unlike K-Means which is hard based grouping. The user can access images from an image search engine, picture library, trained data sets, etc. The images being accessed may have no association with what the user is actually looking for. Hence there is a necessity of providing the user more accurate collection of images which can be done through fuzzy K Means clustering.

SUMMARY This paper deals with the design and implementation of a visual kinematic control scheme for a redundant manipulator. The inverse kinematic map for a redundant manipulator is a one-to-many relation problem; i.e. for each Cartesian position, multiple joint angle vectors are associated. When this inverse kinematic relation is learnt using existing learning schemes, a single inverse kinematic solution is achieved, although the manipulator is redundant. Thus a new redundancy preserving network based on the self-organizing map (SOM) has been proposed to learn the one-to-many relation using sub-clustering in joint angle space. The SOM network resolves redundancy using three criteria, namely lazy arm movement, minimum angle norm and minimum condition number of image Jacobian matrix. The proposed scheme is able to guide the manipulator end-effector towards the desired target within 1-mm positioning accuracy without exceeding physical joint angle limits. A new concept of neighbourhood has been introduced to enable the manipulator to follow any continuous trajectory. The proposed scheme has been implemented on a seven-degree-of-freedom (7DOF) PowerCube robot manipulator successfully with visual position feedback only. The positioning accuracy of the redundant manipulator using the proposed scheme outperforms existing SOM-based algorithms.

This paper is concerned with the inverse kinematic control of a 6 DOF robot manipulator using visual feedback. Two different frameworks have been proposed to learn the inverse kinematics of the manipulator. In the first framework, the robot work-space has been discretized using a priori fixed number of fuzzy regions. Within each fuzzy region, the inverse kinematic relationship from image plane as observed by two fixed cameras to joint space of the manipulator is expressed as a linear map using first order approximation. This proposed framework allows the inverse kinematics to be represented by a Takagi-Sugeno (T-S) fuzzy model whose parameters are learned on-line using gradient descent algorithm. In the second framework, the robot workspace in image plane is discretized into a number of clusters whose centers are determined using Fuzzy C Mean (FCM) clustering algorithm. The FCM algorithm allows each data vector to belong to every cluster with a fuzzy truth value between 0 and 1. The inverse kinematics problem is solved without using any knowledge about orientation of the manipulator. This leads to redundant solutions in the joint angle space for a given target position. This redundancy in the joint angle space is achieved using the concept of sub clustering in the joint space. Inclusion of sub-clustering also improves the position tracking accuracy. The proposed algorithms have been successfully implemented on a 6 DOF PowerCube manipulator from Amtec robotics with a reasonable position tracking accuracy.

This paper presents an elegant method for controlling nonlinear systems by modeling them in terms of a Takagi-Sugeno(T-S) fuzzy model. The concept of network inversion is used to design the controller for such a system. The proposed controller is shown to make the closed loop system stable in the sense of Lyapunov. The existing controller design techniques for T-S fuzzy model, like LMI techniques, robust control techniques are based on a sufficient or prerequisite condition for closed loop stability whereas in the present scheme no such sufficient condition is necessary. Moreover the present approach greatly simplifies the process of controller design compared to the earlier techniques. Simulation results on three nonlinear systems show the efficacy of the proposed control scheme. The proposed controller has also been implemented on the cart pole system in real time and the results are provided with a qualitative comparison with the well established LQR control.