Computer Vision-Based Terrain Sensors for Blind Wheelchair Users
By Coughlan, James; Manduchi, Roberto; Shen, Huiying; ICCHP 2006 - 10th International Conference on Computers Helping People With Special Needs,Publication Date: 2006
Paper describes the development of computer vision technology for sensing important terrain features as an aid to wheelchair navigation. These features include drop-offs, curbs and curb cuts, and the edge of the sidewalk bordering grass or other terrain, or adjoining a wall, known as the shoreline. Computer vision algorithms are used for interpreting visual scenes to infer this visual information, in real time, obtained from images collected by video cameras mounted on the wheelchair. This information is communicated to the traveler using synthesized speech, audible tones and/or tactile feedback, and is meant to augment rather than replace the information from existing wayfinding skills. In experiments with the technology, one or more pairs of stereo video cameras are mounted on the wheelchair and connected to a laptop computer to produce range estimates throughout the image, using the Videre Design MEGA-DCS color stereo system, which produces range maps at several frames per second. In order to find important terrain features, the ground plane in the scene is first estimated, and this information is used to convert the range map into an elevation map that indicates the height of points in the scene relative to the ground plane. Future plans for refining and testing the algorithms for finding important terrain features are discussed.
Published by:
Johannes Kepler Universitaet (Web Site: http://www.jku.at/ )
This publication is included in the library of the National Rehabilitation Information Center (NARIC), accession number O18063
