Department of Computer Science & Engineering

&ldquo;IIT Jodhpur is committed to impart quality education to facilitate the creation of qualified, competent and responsible human resources to meet the emerging technological challenges of the world. It is working towards creating an equitable learning platform that is positive, accessible and effective in the face of the modern societal and scientific advancements.

Flexibility in curriculum, in all its academic programs, encompassing elements of teaching, research, creativity, innovation and entrepreneurship is the most unique feature of IIT Jodhpur. The Department of Science and Technology (DST), under the National Mission on Cyber Physical Systems, has identified IIT Jodhpur as a Technology &amp; Innovation Hub in the area of Augmented Reality and Virtual Reality. The institute has also been nominated as the coordinator for the Jodhpur City and Knowledge Cluster by the Office of Principal Scientific Adviser, Government of India.

The sprawling 852-acre IIT Jodhpur campus is a majestic tribute to the deep-rooted diverse cultural heritage of the region. The campus is an architectural masterpiece, which is slated to become an international exemplar of sustainability with net-zero energy, and advanced water and waste management strategies.&rdquo;

IIT Jodhpur

A query based learning algorithm is proposed in this paper to obtain a valid neural emulator of the robot manipulator using radial basis function networks. This algorithm is centred around the concept of network inversion for which we have proposed an extended Kalman filtering (EKF) based algorithm. Initially the training data set is generated by tracking various random trajectories in robot work space. During this process, control input (or joint actuating torque) vector is computed using a fixed PD control scheme with added dither signal. The RBF network thus learned is tested for its validity using the inversion based recall process. In case the neural network is partially trained, additional input/output data pairs are generated. The data so generated are used to retrain the RBFN model of the robot manipulator. Experiments showed significant improvement in the neural model after retraining. © 1996 Taylor & Francis Group, LLC.

World University root node

IETE Journal of Research

A Query Based Learning Algorithm for Modelling Robot Manipulator Dynamics using Network Inversion

Fast and accurate trajectory tracking of a robot arm primarily depends on the knowledge of its explicit inverse dynamics model. On line learning of inverse dynamics using supervised learning algorithm is difficult in the absence of a priori knowledge of command error. On the other hand, self-organizing neural network employing unsupervised learning scheme does not depend on the command error. These networks are suitable for both off-line and on-line schemes of learning the inverse dynamics. The present paper proposes both the schemes based on two unsupervised learning algorithms, namely, Kohonen's self-organizing topology conserving feature map and 'neural gas' algorithm. Simulation results on a single link manipulator confirms the efficacy of the proposed schemes.

IEEE/IAS International Conference on Industrial Automation and Control, Proceedings

Self-organizing neural networks for learning inverse dynamics of robot manipulator

The paper investigates the application of inversion of a radial basis function network (RBFN) to nonlinear control problems for which the structure of the nonlinearity is unknown. Initially, the RBF network is trained to learn the forward dynamics of the plant. Two different controller structures are then proposed based on this identified RBFN model. In one scheme, a feedback control law is derived based on the input prediction by inversion of the RBFN model so that the system is Lyapunov stable. The second kind of controller structure predicts the feedforward control action, while the fixed controller actuates the feedback stabilizing signal. An extended Kalman filtering based algorithm is employed to carry out the network inversion during each sampling interval. Two examples are presented to verify the proposed scheme. Simulation results show that the performance of the controller based on the proposed network inversion scheme is efficient.

IEE Proceedings: Control Theory and Applications

Inversion of RBF networks and applications to adaptive control of nonlinear systems

A new neural network strategy for clustering is presented. The network works on the histogram and the process is similar to mode separation. The number of clusters are autonomously detected by the network and it overcomes some major difficulties encountered by mode separation techniques. Clustering is done by first selecting the prototypes and then assigning patterns to one of the prototypes based on its distance from the prototype and the distribution of data. The network does not employ weight learning and is therefore faster than existing unsupervised learning networks. The network was applied to a wide class of problems including gray level image reduction, color segmentation and remotely sensed image segmentation. The experimental results obtained are promising. © 1994 IEEE

IEEE Transactions on Systems, Man and Cybernetics

A Connectionist Approach for Clustering with Applications in Image Analysis

This paper presents a neural network based control scheme for robot tracking applications. Gaussian networks are used to model both the forward and the inverse dynamics of a robot arm. The feedforward torque is actuated by the output of inverse dynamic mapping while the feedback control law is derived using the linear perturbation model of the identified forward dynamics along the desired trajectory. The gradient based learning algorithm is used and the effectiveness of the proposed scheme is highlighted through simulation studies. © 1994.

Robotics and Autonomous Systems

Trajectory tracking of robot manipulator using Gaussian networks

This paper is concerned with recognition of hand-written and/or printed multifont alpha-numeric Bengali characters. It is assumed that characters are present in an isolated fashion. In the present work characters have been represented in terms of the primitives and structural constraints between the primitives imposed by the junctions present in the characters. The primitives have been characterized on the basis of the significant curvature events like curvature maxima, curvature minima and inflexion points observed along their extent. Curvature properties have been extracted after thinning the smoothed character images and filtering the thinned images using a Gaussian kernel. The unknown samples are classified using a two-stage feed forward neural net based recognition scheme. Experimental results have established the effectiveness of the technique. © 1993.

Pattern Recognition

Bengali alpha-numeric character recognition using curvature features

This paper presents an abductive formalism for interpreting images of multi-object scenes in terms of objects expected in the scenes. The inferencing scheme provides a mechanism for integrating relational feature-based evidence with the information about the local region properties like color, texture, etc. for making correct recognition decisions. The reasoning scheme is robust enough to tackle the problem of occlusion. A formal definition of the class of plausible interpretations of an image is provided, and competing interpretations are characterized. A problem solving scheme for generation of plausible interpretations has been presented. The concept of the best interpretation of the image has been introduced, and an algorithm has been designed that is guaranteed to provide the best interpretation among a set of plausible conditions. © 1993.

Information Sciences

Abductive formalism for two-dimensional object recognition

A technique is presented for aggregating edge points into polylines which can effectively represent object contours. The technique is based on a new formulation of Hough transform (HT) for detection of line segments. The space requirement of the HT is brought down by considering a different parameterization of straight lines. In this method, the process of edge linking and boundary approximation are combined into a single algorithm. Consequently, the scheme is computationally more efficient than the classical boundary approximation techniques which require a separate edge linking algorithm for preparing their input-the linked list of edge points. Experimental results presented here highlight the effectiveness of this method for approximating object boundaries of polygonal as well as curved shapes present in the images of complex multi-object scenes even in the presence of noise. Further, the parallel algorithm proposed herein for implementation of this technique makes it an ideal choice for real time applications. © 1993.

A new approach for aggregating edge points into line segments

A connectionist model for learning and recognizing objects (or object classes) has been presented here. The learning and recognition system uses confidence values for the presence of a feature. The network can recognize multiple objects simultaneously when the corresponding overlapped feature train is presented at the input. An error function has been defined and it is minimized for obtaining the optimal set of object classes. The model is capable of learning each individual object in the supervised mode. The theory of learning is developed based on some probabilistic measures. Experimental results have been presented. The model can be applied for the detection of multiple objects occluding each other. © 1993 IEEE

IEEE Transactions on Neural Networks

A Connectionist Model for Category Perception: Theory and Implementation

The group of classical graph-theoretic problems, including graph colouring, clique cover, and maximal clique, may be viewed as instances of a general node partitioning problem (NPP). A wide variety of real life problems can be modelled as instances of NPP. Finding an optimal partition for the NPP is said to be NP-complete. In this work a stochastic search by a genetic algorithm (GA) is employed to find a near optimal solution for the NPP. The critical aspects of the GA for NPP, such as the solution representation by elegant data structure, together with genetic operations and selection policies employed in the GA procedure, are also presented. The proposed GA does not require the number of disjunct node sets to be given a priori. Three application problems is VLSI design are solved as instances of NPP. The experimental results presented in each case of these application problems bring out the efficacy of genetic algorithms.