Encoded Visible Light Based Indoor Localization and Navigation Technology for Mobile Autonomous Robots
Reliable autonomous robot navigation is critically important when a mobile service robot is used in a domestic or industry environment. To guarantee an accurate and reliable navigation performance, many practical applications leverage artificial landmarks. In this dissertation, we introduce a visible light-enabled indoor localization system that relies on unique spatial encoding produced when the mechanical mirrors inside a projector are flipped based on gray-coded binary images. Then we present the first encoded projection based navigation system for indoor autonomous robots. At the same time, off-the-shelf photodiodes are used as landmarks and strategically placed on the ceiling to form a topological representation of the environment. The proposed technology makes use of a topological map for global path planning and encoded projection based location discovery for local smooth navigation. With this combination, the proposed scheme is efficient, scalable, and can be applied to mobile robots with limited computation resources in a large-scale workplace. Through experiments within real-world environment, we demonstrate that our proposed approach can locate a target device with an average accuracy of 1.7 millimeters and allows robust autonomous robot navigation in practice with a navigation error of 18.5 millimeters. Furthermore, for dealing with the generally NP-hard multi-robot coordination problem, we discretize the environment into a topological graph and take advantage of its subgraph structure to guarantee the completeness and reliability for real-world multi-robot applications.