Abstract
There are many persons who lost their sight and want to get more flexibility for their activities. But there are very small chances to get new guide dog by about 10%. So we started to serve a guide-dog robot for sight-handicap persons. In the first paper, algorithm of obstacle detection with image processing is proposed. We classified obstacles into three main categories "Pole", "Bar" and "Step" shape, and propose new parallactical difference method included stereo camera on moving vehicle, which detects pole and bar shape obstacles under in-house condition successfully.
