An Autonomous Drone for Image-Based Inspection of Bookshelves

Abstract

This article introduces the UJI aerial library robot. It uses visual techniques to self-locate and navigate autonomously to find books and make automated inventories. A control strategy for navigating along library shelves is presented, using visual markers for self-positioning. An image-based book recognition technique is described, which combines computer vision techniques to detect the labels on the spines of the books, and then optical character recognition (OCR) to decode the book code into text. These data can be used to make an inventory of the library. Lost books can be detected automatically and a particular book can be located in the library. Our quadrotor robot was tested in a real library with promising results.