Vision-based obstacle avoidance for a small, low-cost robot

This paper presents a vision-based obstacle avoidance algorithm for a small indoor mobile robot built from low-cost, and off-the-shelf electronics. The obstacle avoidance problem in robotics has been researched extensively and there are many well established algorithms for this problem. However, most of these algorithms are developed for large robots with expensive, specialised sensors, and powerful computing platforms. We have developed an algorithm that can be implemented on very small robots with low-cost electronics and small computing platforms. Our vision-based obstacle detection algorithm is fast and works with very low resolution images. The control mechanism utilises both visual information and sonar sensor’s measurement without having to fuse the data into a model or common representation. The robot platform was tested in an unstructured office environment and demonstrated a reliable obstacle avoidance behaviour.