Pick’n’place Robot CoWorker In the future, your new apprentice may very well be a robot. A robot that understands your commands and relieves you of tedious repetitive tasks, after brief instruction. A robot that delivers top quality work faster than its human counterpart, takes up less space, and costs a lot less than a 2014 traditional robot system. Sounds too good to be true? The new robot systems that are currently being developed as part of the SMErobotics project promise all that --and DTI in Denmark are on the forefront of this development with their Robot CoWorker platform. Small-batch vision Picture this: Your old-fashioned tool wall. There they all are: hammers, saws, the selection of screwdrivers, … a drill, adjustable wrench, and other essential bits and bobs. All with their own neat silhouette to ensure that every tool finds its way back to its rightful place -- or is immediately visible if not. If you are holding a nail, you instinctively reach for the hammer – it does not matter whether you have one or 50 types of nails, the hammer is always the tool of choice for this sort of job. Every tool on the wall is a specialist. Picking the right tool to get the job done efficiently is a sign of experience that good handymen and seasoned tradesmen have in common. Give a good tradesman a good set of tools and he can do almost anything. Now imagine a robot with its very own tool wall and the skills to go with it. Not literally a wall, but still the general principle is the same. The tools may look different, and the skills may be unlike those of the tradesman, but the overall idea is the same. Select the right tool for the job, get the job done efficiently, and move on to the next job. Traditionally small batch production has not been automated. It has been too expensive. But that is set to change with the new developments in easily trainable CoWorker systems, capable of learning new tasks so quickly that even only 10 repetitions of a task are still worth teaching. And it almost never forgets. But who should train the robot? Not the suit who knows nothing about the job, but the worker who actually performs the manual task normally. The idea is that a skilled worker should do the skilled work, assigning tedious and repetitive work to his robotic assistant. So when a robotsuitable task needs doing, the skilled worker fetches his Robot CoWorker (or takes the task to the robot), chooses the right tool, and teaches the robot what to do. After instruction, the robot can perform the task unsupervised. The robot even learns from its own experience, so it works faster when it “knows” the task. Next time the same task needs doing, the robot doesn’t even need instruction, it recognizes the task for itself, calibrates itself, and starts working. If something doesn’t seem right, it asks its supervisor for help or clarification, as a human trainee would. If something or someone is in the way, the robot works around it or turns to another task -- or waits until the scene is clear again and it can resume its original task. And what one Robot CoWorker knows, they all know. The Robot CoWorker makes it possible for the skilled worker to use a robot in much the same way as he would use an apprentice: to do the less skilled jobs, and learn. To be an intelligent third hand, to assist in assembly, or whatever pick-and-place task that the worker may want to have completed -- while he attends to work requiring more of his skill. In short, the Trainable Robot CoWorker system is your most useful, flexible and versatile tool. You place your robot anywhere you have a suitable task, regardless of the specific properties of the task. Either the robot will start work, or it will ask for clarification and training. Therefore, any small or medium enterprise may have just one Robot CoWorker performing a whole variety of tasks around the building. Versatility A task such as hammering in a nail has many components: Pick up a nail, place it, keep hold, then hit the nail with the hammer – several times. A traditional robot system can be programmed to do all that, accurately and reliably and identical each time. But nothing else. Instead of a dedicated system, optimized for one specific and complete task, our Robot CoWorker system is equipped with many separate skills. Like a person, these simple skills can be combined as needed, so the robot can do a variety of tasks. Maybe not as fast as a dedicated machine, but all the tasks in one little robot. Again and again, to a high standard. Without tiring. Like a human, the Robot CoWorker uses tools. It comes with some of its own, but it can also use other machines. So it can feed the laser scanner, or remove valves from the input tray, adjusting itself to object dimensions, door heights and other variations from workstation to workstation. This versatility makes integration of the Robot CoWorker in any industrial setting less complex than a traditional system. Due to our skill-based modular approach being hardware-independent, our Robot CoWorker system can complement different types of articulated robot, regardless of brand and model. Your business will invest in one Robot CoWorker system and add modules according to your specific needs. One module may be vision, and another a power screwdriver. The possibilities are virtually limitless. This not just economically sane in itself, but also reduces the amount of (expensive) space traditionally required for automation -- another goal of SMErobotics. Current status DTI and the system-integrator BILA are currently working on a case for Danfoss Power Electronics, assembling frequency inverters which are currently assembled manually due to the large variation in client-specific requirements. Robot CoWorker will solve the complete task of picking the various parts and assembling these into the initial subproduct of a frequency inverter. The solution includes the use of advanced vision and force-sensors, in order to pick parts directly from pallet boxes, assembling these in a dedicated fixture, and subsequently mounting screws to fasten the parts together. When the solution is implemented successfully, Danfoss Power Electronics will be able to meet any client requirement, without manual interference or costly setup time. Innovation potential The principles behind any CoWorker system rest on four pillars: Smart setup, Easy instruction, situation assessment and safe execution. Smart setup Every time the physical tool is changed on Robot CoWorker, the system automatically adapts its software-skills to the new hardware. For example, if a new camera is added, the corresponding vision skills automatically become available to the system and accessible to the user. And the user will not have to know anything about the software making this possible. Easy instruction Each assembly line worker knows one specific task, but combined they produce complex objects like the first Model T Ford. In the same way, the simple skills of the Robot CoWorker can be combined to perform semi-complex tasks. These semi complex tasks, put in sequence, complete more complex tasks. Currently skills are being managed and connected on a tablet computer, and the system then translates the skill sequence into complex code executed by a system controller. In the future this will be done even more intuitively. Today free-drive instruction is possible, i.e. the robot is moved manually to show it a path or a position from which to pick or place parts – vision modules will correct these to ensure more optimized flow and successful grasps in the subsequent production runs. Eventually the Robot CoWorker may be voice-instructed and corrected or it may learn a task just by watching the skilled worker, much like a human apprentice. It could be assisted by a “Robot-Google”, where the Robot CoWorker system looks up solutions for specific tasks in a robot process library…. Situation assessment As humans work on a repetitive task they will immediately stop and investigate if they notice something wrong. In the same manner, the Robot CoWorker system learns what is normal during the instruction phase. During execution the robot will stop and inform its supervisor if it notices unexpected sensor input that could indicate problems. The supervisor may label this sensor input as normal and therefore acceptable in the future, or may note this as an anomaly and unacceptable -- and afterwards locate and solve the causal problem. With built-in situation assessment the detection and handling of errors requires no expert resources for programming and adds to the flexibility and positive business case for the Robot CoWorker. Safe execution For our Robot CoWorker system, as for all commercial machinery, safety is the primary concern. Lack of other safe options confined early robot systems inside heavy cages. Modern lowerpowered robots are intrinsically safer, and alternative fenceless safety solutions are available e.g. light curtains, or laser scanner-based zoning of the robot’s work area. Light curtains will stop the robot if the beams are broken, whereas zoning systems allow a more controlled stop, slowing the robot according to the proximity of the intruder. The core idea of CoWorker technologies is the ability to work alongside humans, so humans have to be allowed near. Restrictions on robot operation have to be kept to the minimum that is safe, even when the robot’s workspace is compromised. The Robot CoWorker system integrates that safety into all levels of design. It intelligently re-plans its motion to avoid any obstacle in the original trajectory, or will wait a moment to allow something to move past, and will even keep itself busy with other tasks if its primary task is not currently possible. As soon as there is a free working path, the robot will resume work. So a person can walk through a room of working robots in complete safety, the work around him slowing if he gets too close and resuming when he is out of the way. It will still be a while before the perfect personalized robot assistant familiar from science fiction is a reality. However the Robot CoWorker is one step in this direction. Individual story from the partner Founded in 1906, the Danish Technological Institute occupies a crucial position at the point where research, business, and the community converge. The Institute’s mission is to promote growth by improving interaction and encourage synergy between these three areas. It has around 1000 employees, making it one of the world’s most significant private institutes to supply approved technological services such as consultancy, tests, certification and training for companies and public-sector organizations. DTI adopts an interdisciplinary approach to innovation and to the task of improving the ability of small and medium-sized companies to exploit new technologies and management tools. It has a turnover of some 83 million Euros and is financially sound. 70% of DTI’s customers are SMEs. DTI’s main activities in robotics are located in Odense (Centre for Robot Technology). DTI is actively involved in many robotics projects in industry, education and healthcare as a research partner as well as being involved in dissemination and technology transfer. We have considerable experience in pushing the boundaries with robotic automation to meet actual needs on the factory floor. Therefore we work a lot with error recovery and exception handling, develop functional user interfaces and run training courses for industry. The future will require more flexible solutions with increasing demand for intelligent sensors -- DTI is already working in this area and will be at the forefront with this technology also. (Parts below are fixed and the same for all articles.) The consortium behind SMErobotics The SMErobotics initiative, www.smerobotics.org, consisting of major European robot manufacturers and leading research institutes, is developing and maturing technical foundations for productive and intelligent robot solutions that are feasible in small-to-medium-scale production settings and will showcase some of these at AUTOMATICA 2014. We strive to bring cognitive robotics to the shop-floor, facilitating the use of automation and robot technology to improve the competitiveness of small and medium-sized enterprises (SMEs) in assembly, woodworking and welding scenarios. The project will showcase the following exhibits: Smart Dual Arm assembling (Partner/Robot: COMAU RML) (Link to local project site) Woodworking robot (Partner: Güdel, ULUND; Robot: ABB, Güdel) (Link to local project site) Compliant robot assembly (Partner DLR; robot: KUKA) (Link to local project site) CoWeldRob (Partner: IPA, Robot: Reis) (Link to local project site) DTI Robot CoWorker (Partner: DTI, robot: COMAU) (Link to local project site) Fact Box European open partnership Leading European robot manufacturers Comau, Güdel, KUKA and Reis and internationally renowned universities and research establishments Fraunhofer IPA, Lund University, the Danish Technological Institute, the fortiss Institute at the Technical University of Munich (TUM) and the DLR Institute for Robotics and Mechatronics have come together in SMErobotics to provide access to the technical know-how required for ambitious solutions and genuine breakthroughs in robotics technology. Coordinated by Fraunhofer IPA, a leading institution for applied research, SMErobotics is engaged in close collaboration with various SMEs (e.g. MIVELAZ, WEBER), which will test the newly developed technologies under real-world conditions in four technology exhibits during the project lifetime. The SMErobotics initiative is open to other SMEs wishing to contribute their experience and to benefit from the latest technological advances. Project acronym: SMErobotics. Project full title: "The European Robotics Initiative for Strengthening the Competitiveness of SMEs in Manufacturing by integrating aspects of cognitive systems" Grant Agreement Number: FP7 – IP 287787 SMErobotics partners: Fraunhofer-Institut für Produktionstechnik und Automatisierung (IPA) COMAU Robotics S.p.A. DLR – Deutsches Zentrum für Luft- und Raumfahrt e.V. DTI – Danish Technological Institute GPS Gesellschaft für Produktionssysteme GmbH GÜDEL AG KUKA Laboratories GmbH Reis GmbH & Co KG Maschinenfabrik fortiss – An-Institut der Technischen Universität München Lund University – ULUND For specific questions related to SMErobotics do not hesitate to contact us: SMErobotics Project Secretariat Thilo Zimmermann and Björn Kahl Phone: +49 711 687031-42 secretariat@smerobotics.org www.smerobotics.org GPS Gesellschaft für Produktionssysteme GmbH www.twitter.com/SMEroboticsEU Nobelstr. 12 www.facebook.com/SMEroboticsEU 70569 Stuttgart www.youtube.com/user/SMEroboticsEU Germany