SCS Undergraduate Thesis Topics
|Xinghao Pan||Dave Touretzky||Object Recognition Tools for Educational Robots|
SIFT (scale-invariant feature transform) features, developed by David G. Lowe, have been found to be robust to translations, rotations and scaling, and have become the solution of choice for many when dealing with problems of robotic vision and object recognition. The SIFT algorithm extracts significant features from an image, but additional software is needed to match these features against an image library in order to do object recognition. Further software is needed to construct and maintain this library. At present there is no open source tool to conveniently perform all these functions. This research aims to develop SIFT algorithms into tools that facilitate robotic object recognition for students in undergraduate robotic programming courses. This would involve allowing programmers to use and also to understand the basics of the algorithms behind the tool, and to make adjustments within the object recognition tool to accomplish their goals. As part of the research, the tool will also be evaluated within a real classroom setting when it is released for use in an undergraduate cognitive robotics course.