Visual orientation in the sewer - adaptation to the environment

摘要

Most biological systems employ visually acquired information for their locomotion. In the course of evolutionary history, the visual system of organisms has evolved to be adapted to the environment. As a consequence of this adaptation, biological systems often display highly efficient visual skills. This reasoning has motivated the development of a specific visual system, which serves the purpose of navigation in an unusual environment - a sewer. The sewer environment exhibits two dominating features: restricted geometry of its inner surfaces and absolute darkness. These features are exploited by the hybrid vision system of the autonomous robot consisting of a crosshair laser projector and a camera. If a priori knowledge about the sewer geometry is taken into account, orientation of the robot can be derived from a visual analysis of a regular laser pattern projected onto the sewer surface. Because the footprint image is acquired in an entirely dark environment, the camera records a mostly dark image with the bright footprint in it. The analysis of such an image is very fast and knowledge of the robot's instantaneous orientation derived from this analysis is enough to guide its navigation. It is concluded that proper exploitation of the environmental constraints has lead to the development of this highly efficient visual system.