Active motion vision for distance estimation in autonomous mobile robots

Abstract

Estimation of distance to nearby landscape features is an essential navigation capability for an autonomous mobile robot. The ability to track targets in real time, combined with the concept of motion vision, allows a robot to rely on vision-based navigation techniques and provides an alternative to using active sensors or sophisticated stereo imagers for distance measurement. In this work, we present a real-time vision system capable of estimating distance to objects by tracking their apparent motion, which relies on adaptive temporal motion scaling and optimized tracking algorithms. The vision system is designed for implementation in a chip-level parallel computer architecture and synchronous work with an imager at very high frame rates. The test results demonstrate low sensitivity to noise and interference.