site search  
 
 

PROJECTS

 
 
 









AIBO® Obstacle Avoidance Maze
Fall 2004, Carnegie Mellon University

The goal of this project is to test our cumulative knowledge of the Sony AIBO® Robots and the CMRoboBits source code by having to program a pair of AIBO's to cooperatively make their way through a maze. The maze contains an obstable that the dog will have to climb over and another obstacle to crawl under. At the end of the room, there will be a colored ball pit where the robots will have to cooperatively empty. There are 2 mazes, whose walls are identical but with different features. Two robots will be started in the beginning of each maze, and they will simultaneously navigate through the mazes to reach the ball pits. Each maze set will contain unique landmarks that only one dog will be able to see. Therefore to be able to allow the other dog to navigate in a similar way, the 'observer' will have to communicate with the other dog wirelessly to convey the correct information at the right time for synchronized navigation.


Proceed to details

 

Augmented Reality
Fall 2004, Carnegie Mellon University

Augmented Reality has become more and more popular since computing powers are offering almost unlimited possibilities. Augmented Reality is a special form of Virtual Reality which is mainly the overlay of computer generated images onto the physical environment. Take movies nowadays, which make great use of 'special effects' to inject virtual objects in the video sequences to render movies more appealing. This can be easily done by using special markers while shooting the movie, and the video stream can be either manually or automatically edited to inject the virtual objects into the movie. For my project, the challenge was to create Augmented Reality in an uncallibrated video sequence. This is a totally different story when the video stream does not contain any special markers that can be edited.

Proceed to details


 

Template Motion Tracking in a Video Stream
Fall 2004, Carnegie Mellon University

Motion Tracking has been one of the hot topics of research in the field of Computer Vision for a while. This is due to the wide varieties of applications that could result from such research in the areas of surveillance, military, guidance systems and many more. In this project, I will be using a specific technique of motion tracking namely, template tracking. For this particular system, a region of the video feed is manually selected, unless a distinct template is preloaded to the system. This selected region is then used a template to be tracked throughout all the subsequent frames of the video feed.

Proceed to details

last 3 << >> next 3



:: Site Map :: Contact :: Projects
©2007 Kenny Teng. All rights reserved