Perception-Guided Mobile Manipulation Robots for Automation of Warehouse Logistics

Dr. Moritz Tenorth is CTO of the Munich-based startup Magazino that develops perception-guided mobile manipulation robots for the automation of warehouse logistics. The company specializes on the development of robots that pick and place individual products on shelves for book distributors and e-commerce shoe vendors. Dr. Moritz Tenorth acquired his PhD at Technische Universität München (TUM), Germany, in 2011. From 2011 until 2014, he worked as a Post-Doctoral researcher at TUM, the Advanced Telecommunications Research Institute International (ATR) in Kyoto, Japan, and at the University of Bremen, Germany. In 2015, he joined Magazino first as Head of Software Development, and as of 2017 works as Magazino’s CTO. His research interests are in the fields of knowledge representation and reasoning for robots, as well as cloud robotics. Künstliche Intelligenz (KI) spoke with Dr. Moritz Tenorth by correspondence in August 2018. The question set was compiled by Georg Bartels and Michael Beetz. KI: Moritz, Magazino has achieved a lot of initial success as a German robotic startup. Considering the high labor costs in Germany and technology-intensive nature of robotic products, which particular challenges are you facing, compared to more “traditional” software and business startups? With Magazino, we are among the first to bring a perception-guided mobile manipulation robot out of the laboratory into real applications. First of all, our robots are highly complex technical products, and being able to manage this complexity in design, programming, production and operation is a challenge of its own. In addition, the area of autonomous robots is a very new one where little is known about best practices to develop an actual product – how to engineer a control system for a perception-guided autonomous robot, how to debug it, how to improve performance, how to ensure robust operation? This is different for instance in comparison to Web startups that can build upon a vast range of infrastructure, libraries, proven approaches and best practices; we have to develop many of these ourselves. While an iterative approach to development, as common in most startups, still makes sense of course, each iteration costs significant time and money since hardware has to be ordered, produced and can only months later be tested in the real environment. KI: Magazino is a company with a strong research background that tries to establish cutting-edge technologies in the market. What challenges are you experiencing in following this particular path? Ensuring sufficiently robust operation, as expected from a product, for this very new technology is challenging since the environment the robots operate in is only * Georg Bartels georg.bartels@cs.uni-bremen.de