This was the result of my masters thesis project at UCSB. In this demo LittleDog uses a webcam mounted on its body to capture snapshots of upcoming terrain and chooses its gait to maximize speed. This system does not rely on motion capture and uses a mid-range laptop for real time terrain classification. This project relies heavily on open source projects including OpenCV, wxWidgets, and QT. The visual recognition code for this work can be found here.
iOS Object Recognition
This was done for a project in one of my computer vision courses. It demonstrates server-side image recognition with an iOS device. In this demo the iPod Touch (4th gen) takes several pictures of app icons from my computer monitor using the built-in camera software. My app, nicknamed cAPPture, then crops/scales each image and ships it off to an image recognition server over wifi. The server matches the image to one of its database images and sends the corresponding name back to the iPod in real-time.