An active vision architecture based on iconic representations
Artificial Intelligence - Special volume on computer vision
Hi-index | 0.00 |
Estimation of distance to nearby landscape features is anessential navigation capability for an autonomous mobilerobot. The ability to track targets in video sequences,combined with the concept of motion vision, allows arobot to rely on vision-based navigation techniques andavoid using active sensors or sophisticated stereo imagersfor distance measurement. We present recent developmentsof our work on a real-time vision system capableof estimating distance to targets by tracking theirapparent motion. The algorithm is designed for implementationon chip-level parallel computer architectureand to work synchronously with an imager at very highframe rates. Test results demonstrate sufficient stabilityof tracking and low sensitivity to noise and interference.