Free Space Detection from Catadioptric Omnidirectional Images for Visual Navigation using Optical Flow

In this paper, we develop a free space detection algorithm for the visual navigation of the autonomous robot mounting a catadioptiric omnidirectional imaging system. The algorithm detects the dominant plane as the free space from a sequence of omnidirectional images captured by a camera mounted on the autonomous robot. The dominant plane, which can be detected from the optical-flow field, is the largest planar area in the image. For the detection of the dominant plane from the optical-flow field, we adopt the motion separation property, that is, the optical-flow vector is decomposed into infinitesimal rotation, translation, and divergent motions on the images. The algorithm matches the measured translation optical-flow field with the template translation optical-flow field to separate the dominant-plane as the free space for the navigation and the obstacle area. The template optical-flow field is generated from a preobserved image sequence without any calibration of the internal parameters of both the robot and camera.