Hybrid field of view vision: From biological inspirations to integrated sensor design

Abstract

In this article a hybrid field of view vision system employing both a catadioptric camera and a typical perspective-view camera is presented and evaluated. This configuration takes biological inspirations from the peripheral and foveal vision co-operation in animals and compensates the most important drawbacks of both camera types. To make the sensor self-contained and suitable for mobile robots lacking a powerful on-board controller the design is based on a recent generation single-board computer providing parallel processing by means of GPGPU. The software of our sensor permits to use it for quick detection of obstacles in the vicinity of the robot, and to measure distances using stereo vision.