Indoor navigation for mobile robots using memorized omni-directional images and robot's motion

Abstract

We have proposed a navigation method for mobile robots in indoor environments based on a memorized sequence of images and robot's motion. The method consists of two stages: route teaching and playback navigation. At the teaching stage, the robot is controlled to move along a designated route in an environment, records robot's motion, and memorizes environmental images. In the course of autonomous navigation, the robot plays back the recorded data. It compares the currently taken image with the memorized one so that it estimates its position, and then calculates a trajectory to track the memorized route in real-time. We use an omni-directional camera to perceive environmental information, and extract the vertical edges existing in environment as landmarks in this method. In this paper, we report on the use of the color on both sides of vertical edges so as to match these edges more robustly, and explain the autonomous navigation in backward direction using the forward taught data.