Robot navigation behaviors based on omnidirectional vision and information theory

Abstract

Abstract-In this work we present a reactive autonomous robot navigation system based only on omnidirectional vision. It does not rely on any prior knowledge about the environment apart from assuming a structured one, like indoor corridors or outdoor avenues. The direction of the corridor is estimated from the entropy analysis of a 1-D omnidirectional image. The 2-D omnidirectional image is analyzed for obstacle avoidance and for keeping a safety distance from the borders of the corridor. Both methods are non-metric and no 3-D information is needed. The system performs well with different resolutions and the catadioptric sensor needs no calibration. We present results from indoor and outdoor experiments.

Autonomous exploration using rapid perception of low-resolution image information

“…In contrast, when the robot is turned so that it faces one of the side walls, the range of visible depths is much smaller, and therefore the variety of pixel intensities usually decreases. A similar observation has been noted by other researchers in the context of using omnidirectional images [5,11], but we show that the relationship between entropy and orientation holds even for standard camera geometries. In addition, we have found that the relationship is not significantly affected by whether the walls are textured.…”

Extracting Minimalistic Corridor Geometry from Low-Resolution Images

“…The reason for this perhaps surprising result is that such an orientation causes scene surfaces from a variety of depths to be visible, yielding an increase of image information at this orientation. A similar observation has been noted by other researchers in the context of using omnidirectional images [2,3]. We divide the image into overlapping vertical slices and computing the graylevel entropy of the image pixels in each slice.…”

Autonomous navigation and mapping using monocular low-resolution grayscale vision

“…Using entropy (in addition to the existing methods), therefore, is a promising way to react to a situation where the image does not provide enough information for navigation. Other researchers have used entropy for determining the direction of navigation and for global visual localization using omnidirectional images [6,16]. Entropy is used in several ways.…”