H : 'H'  was the first robot designed  and built by Vijay.  He wanted to build a humanoid like 'Leutenant commander Data'. But 3 months was a short time. So H became the first step towards a complete humanoid. When completed for the robot design course, H used wheels for locomotion, IR sensors for human detection, bump sensors for moving away from small objects and a single joint arm for grasping objects. H operated in the rotunda of the engineering building following lines on the ground. When H detected a human being on its path, H would stop and raise its arm. If the human wanted to give it something, H would take it with its fingers. When the human moved away, H would continue on its path, reach  a collecting bin and drop the object in it. (More details)

______________________________________________________________________________________________

MEMS Integrated 3-axis accelerometer : A sensor for monitoring the health of babies and elderly people by measuring movements was required. A 3 axis integrated MEMS accelerometer was designed by a team of 4 people.  The design was based on a capacitance type accelerometer using both spring and torsion effect for 3 directions. The electronic circuit was also built on the chip. Simulations were performed using coventorware, spice and simulink. Vijay's main role was in designing the control system for stabilizing the moving parts.     __________________________________________________________________________________________________________

Recognition system for animal identification in wildlife photos 
Wavelet transform was used to obtain a two level wavelet decomposition of wildlife photos. Vector quantization was then used to obtain models of the animal and the background. These models were used to segment animal and background.  A neural network based animal recognition system was also used. This work is reported in NeuroSolutions.      __________________________________________________________________________________________________________

Face recognition: Face recognition system was created using three different methods namely neural networks, geometric pattern matching & local linear embedding. The recognition system identified a person's name based on frontal images of the face. Such a system could be used in companies and airports. The yale face database was used for the experiments. Programming was done in Matlab.

_____________________________________________________________________________________________

 Robotic packaging system: A Fanuc robot arm was interfaced with an Allen Bradley PLC  for a typical factory packaging system. Objects were transported through a chute, where IR  sensors connected to the PLC determined the size of the object. This information was then  communicated to the robot arm, which picked up the object and a placed them in slots on a  packaging box. The project was done by a team of 4. Vijay's main role was in designing the  algorithm and programming the robot arm. 

____________________________________________________________________________________

Role of Arm swing on human walking: Many current research models in biomechanics use  a point model for the upper body analyzing the dynamics of the lower limb movements.  This project uses the full upper body and analyzes the role of arm swing in stabilizing the  body during walking. Simulations were performed using SDfast and SIMM. (Click on pic to see video)

__________________________________________________________________________________________________________