Intelligent Systems Laboratory

Department of Electrical Engineering
Indian Institute of Technology, Kanpur

Research Projects


Traditionally we observe natural systems and then we are inspired to build artificial systems. Systemic theory has its origin in our understanding of natural systems. We take this route to develop innovative ideas to design intelligent systems. Research topics include Intelligent control, Assistive robotics, Cognitive modeling and Quantum learning system.

Intelligent Control


System identification and control of nonlinear systems can be efficiently carried out using intelligent control schemes. In this framework, our research work primarily focus on direct, indirect and model predictive control schemes using T-S fuzzy model, neural and fuzzy-neural networks. These control algorithms are being tested in real-time systems such as 7DOF Power-cube robot manipulator and inverted pendulum.

  1. Prem Kumar P., Indrani Kar and Laxmidhar Behera Variable gain controllers for nonlinear systems using T-S Fuzzy model, (To be published) IEEE Trans Systems, Man and Cybernetics, Part B, 2006

  2. Laxmidhar Behera, Query based model learning and stable tracking of a robot arm using radial basis function network, Computers and Electrical Engineering 29 (2003) 553-573

  3. Indrani Kar and Laxmidhar Behera, Neural network based direct adaptive control for a class of affine nonlinear systems, Accepted in IEEE Int. Symposium on Intelligent Control (ISIC), 2006, Munich, Germany.

  4. Prem Kumar, Indrani Kar and Laxmidhar Behera, Intelligent Control Schemes for a Redundant Manipulator, 2nd Indian Int. Conf on Artilficial Intelligence, Pune, India, 20-22, December 2005

  5. Dip Goswami, Laxmidhar Behera, and Ashish Dutta, Simulations and Experiments on a Robotic Arm, CIRAS-05, 10-14, December 2005, Singapore

Intelligent Assistive Robotics

Visual Motor Coordination


Visual-motor coordination, also referred to as hand-eye coordination, in the context of robotics is the process of using visual information to control a robot manipulator to reach a target point in its workspace. The task requires learning the mapping that exists between camera output and desired end-effector location. Biological organisms have demonstrated their superior adaptive capabilities in motion control over present-day robotic systems. Inspired by this fact, various neural network models based on biological systems have been developed for robot control tasks. We have been working on various self-organizing schemes such as Kohonen SOM, Quantum clustering SOM and parametrized SOM to learn this map using minimum learning example. Simultaneously we are extending these approaches to the redundant manipulator system which involves optimization of multiple objective functions.

  1. Nimit Kumar and Laxmidhar Behera, Visual Motor Coordination Using a Quantum Clustering Based Neural Control Scheme, Neural Processing Letters, Volume 20: 11-22, 2004.

  2. Anjan K Ray, Mayank Agrawal and Laxmidhar Behera, Kinematic Control of Robot Manipulators using Visual Feedback, Accepted in IEEE Int. Symposium on Intelligent Control (ISIC), 2006, Munich, Germany.

  3. L. Behera and Nandagopal K., A hybrid neural control scheme for visuo-motor coordination, IEEE Control System Magazine, vol. 19, no. 4, August 1999, pp. 34-41

Visual Tracking of a Mobile Robot


The work is in progress for visual tracking where various algorithms such as Kalman filter, Unscented Kalman filter, and quantum neural network based algorithms are being investigated for efficient tracking of the moving target by a mobile robot. The experiments are being conducted on Patrolbot.

Neural Networks: Training Algorithms

Research in this area focusses on efficient training schemes that would provide fast convergence, better generalization and least computational effort.

  1. Laxmidhar Behera, Swagat Kumar and Awhan Patnaik, On adaptive learning rate that guarantees convergence in feed-forward networks, (To be published) IEEE Trans Neural Networks, Vol. 17, No. 5 September 2006

  2. Laxmidhar Behera, Swagat Kumar and Subhas Das, Identification of non-linear dynamical systems using recurrent neural networks, TENCON, 2003, Bangalore

Quantum Learning Systems

Computing power has grown at an astounding pace since the synthesis of first transistor in 1947. Gordan Moore in 1965 stated that computing power will double for constant cost every two years, and this statement, which has stood the test of time, has come to be known as Moore's Law. At this exponential speed, it was conjectured that it will not be long before we have extremely small sized computing devices. At such small sizes, laws of classical mechanics are no longer valid. Thus came the idea of Quantum computation in his celebrated paper, but it was not before discovery of algorithm for polynomial time factorization by Shor, that the general interest in Quantum computation started increasing.

Thus topics related to "Quantum Computation" and "Quantum Information Processing" are at the forefront of research interest due to rapid theoritical and experimental advances. Apart from the computational power derived from quantum computer, the field of quantum learning systems is poised to provide a potent framework to study subjective aspect of matter. The challenge to bridge the gap between physical and mental (or objective and subjective or mind and body) notions of matter can be countered within the quantum learning systems. In this framework, notion of "quantum brain" and concepts of "recurrent quantum neural networks(RQNN)" have already been coined by researchers.

First step to develop quantum learning system is to build quantum neural network models for applications for which human brain is very capable of. Some such applications are pattern recognition and language understanding. Quantum dynamics of such learning system can be based using Schroedinger wave equation, Feynman path integral, density matrix approach and linear operator theory in Hilbert space, Thus the field of quantum learning system is pregnant with many novel ideas and possibilities.

  1. Indrani Kar, Richa Tayal and Laxmidhar Behera, Quantum Memory and Pattern Retrieval, Internation Journal of Lateral Computing, Vol. 1, No. 2, 2005

  2. Laxmidhar Behera, Indrani Kar and Avshalom C. Elitzur, Recurrent Quantum Neural Network Model to Describe Eye Tracking of Moving Target, Foundations of Physics Letters, Vol. 18, No. 4, 357-370, 2005

  3. Laxmidhar Behera, Bharat Sundaram and Gaurav Singhal, Speech enhancement using a recurrent quantum neural network, Indian International Conference on Artificial Intelligence, IICAI-2003, Hyderabad, India.

  4. Laxmidhar Behera and Bharat Sundaram, Stochastic filtering and and Speech Enhancement using a Recurrent Quantum Neural Network, Proceedings Int. Conf. Int. Sensors and Inf. Processing, ICISIP-2004, Chennai, 165-170.

  5. Laxmidhar Behera and Indrani Kar, Quantum Stochastic Fitering, IEEE SMC 2005, Hawaii, Oct 10-12, 2005


[ Home ][ IITK home ] | [ EE home ] | [ Robotics center ] | [ IITK Bus Schedule ] | [Check Mail ]