My general research focus is in the field of Computer Vision. I am currently researching the use of object boundary information in object recognition. I am interested in the use of limited 3D information for augmenting current 2D-only recognition approaches, but I hope to avoid the computation, complexity, and potential overkill of full 3D reconstruction/ modeling. To that end, I have developed methods for incorporating object boundary information into feature-based object recognition, resulting in an augmentation of the popular Scale Invariant Feature Transform (SIFT) method which we call the Background and Scale Invariant Feature Transform (BSIFT). Now, we are developing robust and accurate methods for finding object boundaries. We also hope to incorporate shape information in the future.
In the past, I have also explored topics such as real-time tracking, assembling 3D object models via ICP, colored structured light for range imaging, and sequential structure from motion (SFM). I am also interested in the problem of mobile manipulation: building robots capable not only of navigating in an environment, but also interacting with that environment. Applications of computer vision and robotics that interest me include assisting people with disabilities, space exploration, search and rescue, and medicine.