A Novel Vision-based Reactive Navigation Strategy based on Inverse Perspective Transformation

This paper describes a new vision-based reactive navigation strategy addressed to mobile robots, comprising obstacle detection and avoidance. Most of the reactive vision-based systems base their strength uniquely on the computation and analysis of quantitative information. The proposed algorithm combines a quantitative process with a set of qualitative rules to converge in a robust technique to safely explore unknown environments. The process includes a feature detector/tracker, a new feature classifier based on the Inverse Perspective Transformation which discriminates between object and floor points, and a qualitative method to determine the obstacle contour, their location in the image, and the course that the robot must take. The new strategy has been implemented on mobile robots with a single camera showing promising results.