Library holdings are only useful if they’re findable. While many libraries focus on the digital aspects of improving search, for print collections at least, even recommending the most relevant titles ultimately falls short if they’re not on the right shelf. Misfiled materials can lead to major controversies, such as the one that recently led to the resignation of former Boston Public Library president Amy Ryan, as well as that staple of library human interest stories, the rare book or manuscript “discovered” in a library or archive. However, the process of finding out if things have been properly shelved is time-consuming and never ending, as materials are continuously moved even if they don’t circulate outside the building. The task is often handled by support staff, interns, or volunteers, but Singapore’s National Library Board has a new alternative: a library robot. It’s the brainchild of researchers at the infocomm research branch of Singapore’s Agency for Science, Technology and Research (A*STAR), a global research organization with 14 research institutes throughout Singapore. The agency developed the machine in response to the national board, which wanted to improve productivity in shelf reading operations but found no commercial solution available. Instead, a group of robotics, wireless communication, antenna design, data analytics, and visual computing experts came together to create a solution of their own. The result is AuRoSS (autonomous robotic shelf scanning system), a robot that can navigate and scan library shelves for misplaced books. The not-at-all-humanoid robot wheels around library stacks, using RFID and laser mapping technology to scan shelves with 99 percent accuracy. It then produces a report of missing and out-of-sequence books that can be given to a staffer for easy re-shelving.. In developing the robot, researchers faced a few big hurdles. The robot had to be made small enough to get between shelves, but big enough to reach high and low locations. It had to be far enough away to not threaten shelved books, but be able to get close enough to “see” individual books and make sure they were in place. And it had to be able to handle varied library layouts. Eventually, said researcher Renjun Li in a release, “we decided to detect the shelf surface itself, and use that as a reference.” Using the shelf as a guide, the robot is able to successfully navigate even off-sized shelves. Dr. Ho Chin Keong, deputy department head of the Cognitive Communications Technology Department within A*STAR’s Institute for Infocomm Research, told Library Journal that the team was tasked with demonstrating the robot inside a newly-renovated library on a tight schedule. A literal curveball was thrown to the robot with the installation of curved library shelves. Only a few of the shelves were available, so the team had to figure out whether AuRoSS was capable of navigating special shelving without much to work with. “We had to do a public demonstration on the day of re-opening,” recalled Keong.  The robot passed the test, reading the curved shelves at the Pasir Ris Public Library just as well as traditional shelving. Currently, AuRoSS is being demonstrated at night within select Singapore public libraries. The next step, said Keong, is gathering commercial interest for a larger pilot program. For Keong, the most gratifying part of designing a whirring robot shelving assistant is seeing how librarians light up when they realize that the amount of time they need to set aside for such tasks can be reduced. But he assures those worrying about automation reducing library jobs that his invention will never replace librarians themselves. “Definitely not,” he said. “The robot only performs the most menial and time-consuming part of the shelf reading operation.” That leaves librarians free to do what they do best, he said: engage with their patrons.