In this group project, we designed and built a lamp to easily identify veins for drawing blood, for IV administration or for injections. This device is useful when vein identification is challenging and time consuming, specially in patients who are, elderly, dark-skinned, obese, or infants.

VeinLite uses the principle of side-trans-illumination, where light shines into the skin from outside the area of interest of the skin and makes the surrounding skin translucent. The deoxyhaemoglobin in venous blood absorbs light, so veins show up as dark lines against the now translucent skin tissue. We used red LED of 760 nm wavelength (which is more penetrative) to enhance this effect

The purpose of this robot was to identify three types of colored boxes. Robot had to navigate to these boxes, by following a set of lines. Once it reaches to a box, the robot should identify its color and keep count of each type of boxes that it encounters throughout its journey. At the end of path it displays the results on an LCD display.

We used a PIC (PIC 16F877A) microcontroller programmed using MikroC as the central controlling unit, and employed an IR reflector sensor array for line sensing, a sonar sensor to detect the presence of boxes, a custom made color sensing circuit using LDR and LED and 16X2 LCD display to output the count.

The mobile robot was designed for an inter class competition to perform a sequence of tasks within the shortest time duration of time. The robot had to solve and navigate a maze to select and transport objects from the source to the destination.

My contribution was developing an algorithm based on Trémaux's algorithm to initially solve the maze. Subsequently, once the paths are identified, methods of Graph theory were used to identify the closest path.

A semi-automatic interactive algorithm that segments images along borders was conceptualized and implemented from the scratch in MATLAB. This algorithm automatically selects a line that connects two selected points by assigning cost to a connection from one pixel to the neighbour pixel and finding the optimum path from point A to point B such that the cost is minimized. The cost is calculated based on the Laplacian zero crossings, magnitude of gradients and the direction of gradient (sobel). Optimum path was found using Dijkstra’s Algorithm. [Source]