European manufacturing industry faces increasing requests for high product variance, small product series, shorter production cycles, and cost reduction. However, few robotic components are designed for easy adaptation and reuse. R5-COP focuses on agile manufacturing paradigms and specifically on modular robotic systems to over-come the shortcomings of existing solutions. R5-COP starts by identifying and extending suitable existing methods and, where required, developing new ones to formally model hardware and software components. This approach will support model-based design, engineering, validation, and fast commissioning. Furthermore, using existing interface and middleware standards such as ROS, R5-COP will strongly facilitate integration of components from various suppliers.
Such a modular approach will not only be more flexible than state-of-the-art solutions, but will also reduce design, setup, and maintenance costs. As flexible use of robots includes their close cooperation with humans, robustness and safety are crucial requirements which will be assured by dedicated verification and validation methodologies.
Within the first year, R5-COP has already produced visible results in all targeted application scenarios ranging from industrial robots to professional service robots and field robots. Core topics of all these applications are autonomy and adaptability, not only requiring a distinct set of flexible sensors but, depending on the use case, also flexible mechanics.
Already early in the project, the Dutch project partners Alten, Robomotive, and TU Eindhoven teamed up with an external industry partner targeting a particular use case: picking up and identifying irregular shaped items for further processing. This is motivated by the fact that today's household item often enough consist of a multitude of complex-shaped parts. Using conventional processing, such parts would require dedicated machinery with individual grippers as well as particular training for that one job and item. Once the item changes, also the processing chain must be changed. R5-COP targets this issue by introducing ``intelligence'', training the robot system to identify and handle a variety of parts. If handling is unknown, it is learned by demonstration. A similar approach is targeted by combined industrial demonstrator procured by Finnish partners VTT and Profin, dealing with identifying and retrieve desired parts from a warehouse and visual inspection of those parts afterwards before processing them further.
Wooden parts retrieval and inspection platform
Picture: VTT
Such not only requires flexible manipulators, but also a powerful vision and decision system. Such is paramount in the case of unfenced robot use like human-friendly environments or even in environments where humans and robots interact. One example of this are the patrouling robot developed by the Polish partner PIAP, or the inspection robot developed by the Norwegian consortium consisting of SINTEF, Statoil and NTNU. Such systems not only need to find their way through environments autonomously, but must also cope with changing lighting, inside vs. outside conditions, and a variety of terrains ranging from e.g. paved roads to soil and grass and furthermore also need to be prepared for human interference.
Inspection & Patrouling Environment
Pictures: Statoil / SINTEF
Key to such patrouling and inspections systems is a powerful visual system, able to provide required sensory information. This may include stereo vision for deriving depth information, but also specialized visual systems like thermal cameras: the latter is, for instance, required for the farming robot demonstrator procured by Probot and VTT, which uses the thermal information to detect plant diseases such as potato blight.
Crop disease inspection using farming robots
Pictures: Probot
These systems need to fulfill strict safety requirements. One building block is safety by design using formal methods, providing mathematical methods to assess a system's safety. Building autonomously reacting and learning systems, however, will not allow full formalizations, instead, systems also need to be thorougly safety-tested using well-defined procedures. One particular test-bed installation is built by the Czeck project partners BUT in cooperation with CAMEA, which is set up to handle different terrains (indoor vs. outdoor) and different types of robots while operating in a human-friendly environment.
Robots in a human-friendly environment -- Testbed setup
Picture: BUT
Giving feedback to the robot operator also is one safety building block. Using dedicated user interfaces does not only allow status reporting, but also job deployment or emergency situation handling. One example of such is the Search & Rescue UGV developed by Swedish partners Lulea University and Swedish Space Corporation. Built around a stock Toyota Landcruiser, this vehicle allows remote monitoring and interaction through a versatile control UI.
Search and Rescue UGV: Sensors/Actuators (left), Control UI (right
Pictures: LTU/SSC
Similarly, the flexible logistics robot platform of Danish partners DTI and Mobile Industrial Robotics (MiR), not only sports versatile robot hardware, but also a detailed user interface for job deployment and robot control. This platform is currently used in a hospital field test where it serves for waste-disposal and other transport purposes. In order to easily couple with a variety of different payload waggons, it features a flexible coupling facility.
Task-deployment console (left) and Flexible Logistics Platform in action (right)
Pictures: DTI
Two tired but happy heroes: PIAP SCOUT® (left) and MiR 100 (right) posing with their award
Picture: Thomas Almdal, DTI
Two of these platforms, PIAP SCOUT® and the DTI/MiR logistics platform (MiR100), were featured on the project exhibition taking place during the ARTEMIS/ITEA2 Co-Summit 2015. Visitors were able to interact with the autonomously driving MiR100 and had the possibility to control the SCOUT via its dedicated remote interface. Our two robotic heroes surely struck some string with the visitors, being finally rewarded with the 2015 Co-Summit Exhibition Award for their constant (and collision-free!) efforts in human/machine interaction. Thanks to all visitors for their interest and positive voting!
Contact: Rainer Buchty, buchty@c3e.cs.tu-bs.de
Project website: http://www.r5-cop.eu/