A memory-based approach to navigation

Abstract

A memory-based system for autonomous indoor navigation is presented. The system was implemented as a follow-midline reflex on a robot that moves along the corridors of our institute. The robot estimates its position in the environment by comparing the visual input with images contained in its memory. Spatial positions are represented by classes. Memories are formed during a learning phase by encoding labeled images. The output of the system is the a posteriori probability distribution of the classes, given an input image. During performance, an image is assigned to the class that maximizes the probability. This work shows that extensive use of memory can reduce information processing to a simple and flexible procedure, without the need of complicated and specific preprocessing. The system is shown to be reliable, with good generalization capability. With learning limited to a small part of a corridor, the robot navigates along the entire corridor. Furthermore, it is able to move in other corridors of different shape, with different illumination conditions.