Application of Optical Flow Sensors for Dead Reckoning, Heading Reference, Obstacle Detection, and Obstacle Avoidance.

摘要

A novel approach for dead reckoning, heading reference, obstacle detection, and obstacle avoidance using only one optical mouse sensor was presented in this thesis. Odometry, position tracking, and obstacle avoidance are important issues in mobile robotics. Traditional odometry and motion-tracking sensors provide relative displacement data on a frame-to-frame basis, and they are usually mounted in arrays to provide accurate measurements with small estimation errors. Optical flow sensors stand as a tempting solution for robot self-localization and dead reckoning. In this work, using only one inexpensive optical mouse sensor, we were able to perform optical odometry, dead reckoning, and heading reference. Also, obstacle detection and avoidance remains a challenging area of research. Most of the existing works are based on stereo imaging and computation of the time-to-contact. These techniques are complex and usually require the use of more than one vantage point. The use of one optical mouse sensor as an obstacle-detection sensor was proposed in this work. The detection process is simple and is based on the surface-quality factor variation. As far as we know, no one has ever used this technique to perform obstacle-detection and avoidance. Using one sensor for motion tracking and one sensor for object detection in association with an Arduino microcontroller, we built an indoor ground robot capable of environment sensing, obstacle avoidance, and position tracking. The behavior of the robot can be monitored from a remote station. The experimental results obtained were promising and can be further improved.