Novel Active-Vision-Based Motion Cues for Local Navigation

摘要

In the absence of a-priori information about the information an autonomous mobilerobot relies on sensory information to make local judgments about its surrounding. Generation of a local collision-free path based on sensory data plays an important role in the control of the robot's motion. The paper presents a novel active vision-based approach for generating local collision-free paths for mobile robot navigation in indoor as well as outdoor environments. Two measurable visual motion cues that provide some measure for a relative change in range as well clearance between a 3D surface and fixated observed in motion are described. These visual cues are independent of the 3D environment and need no a-priori knowledge about it. For each visual motion cue, there is a visual field surrounding the moving observer. In other words, there are imaginary 3D surfaces attached to the observer that move with it, each of which correspond to a value of the cue. These visual fields can be used to demarcate regions around a moving observer into safe and danger zones of varying degree to make local decisions about the sterring as well as speed commands to the mobile robot. We describe a practical method to extract these cues from a sequence of images. This approach needs no feature tracking between images and almost no camera calibration.