Next Generation Robotics for Norwegian Industry

Next-Generation Robotics for the Norwegian Industry ? was a 5-year research program where we have developed new competence and technology in next-generation robotics for Norwegian industry. The project was led by SINTEF ICT and was funded with support from the Research Council of Norway and from the industrial partners Statoil, Hydro, Tronrud Engineering, Glen Dimplex Nordic, SB Seating and RobotNorge. SINTEF researchers, four PhD candidates, one PostDoc and professors at NTNU (Department of Engineering Cybernetics and Department of Production and Quality Engineering) worked on industry-relevant issues identified in cooperation with the industry partners. The project had a case-based approach where relevant industrial use cases were identified and analysed in cooperation with the industrial partners. The use cases were: ? Remote inspection and maintenance of oil and gas installations. ? Cleaning of carbon anodes. ? Packing of chair parts in cardboard boxes. ? Bin-picking on chair wheels. ? Handling of flexible beams. ? Handling of dynamic objects for hanging conveyers. In order to target the identified use cases, a number of new and enabling technologies were developed, tested and demonstrated in NextGenRob. A key focus has been on compentence and technologies which can enable robotized production and operations in changing environments with changing tasks that need to be carried out. Such operations include, e.g., "one-piece production" which entails production lines that can handle a large range of different product types/sizes/shapes. The main competence building in NextGenRob has been within three areas: Robot sensing encompasses how to automatically perceive the environment. ? 3D object localization: In NextGenRob, we have developed tools for real-time analysis of 3D sensor data to determine the position and orientation of objects (e.g., chair parts, or a sensor in an oil and gas facility) in order to enable robots to automatically handle such objects. ? Motion estimation of flexible beams and free-hanging objects: We have developed tools to determine the shape and position of flexible beams for automated handling. Moreover, new tools to determine and track the position of free-hanging objects based on real-time video images and methods for state estimation in order for robots to automatically interact with the objects have been developed. Robot control is about how to control a robot such that it carries out its tasks in the right manner. ? Synchronized control robot manipulators: Robots that cooperate to solve industrial tasks needs to operate synchronized. We have demonstrated synchronized control of a robot manipulator acting as a camera platform for intuitive and safe monitoring of remote controlled operations. Moreover, we have developed technologies which enable robots to cooperatively lift and handle objects. This is beneficial since, e.g., this leads to a reduced need for specialized robot grippers. ? Real-time control: Robot manipulators can operate at very high speeds, but this often comes at the cost of reduced accuracy. NextGenRob has pushed these boundaries and produced methods for high-speed and high-accuracy robot operations. Moreover, we have designed flexible robot interfaces which facilitate easier programming of robots from various vendors. ? Nano-robotics: The ability to sense and manipulate on a very small scale ? at nanoscale ? opens many opportunities within both industrial and consumer technologies. These possibilities include being able to produce smaller and more accurate sensors and better materials (e.g, stronger, larger heat-resistance, etc.). In NextGenRob, we have developed new methods for manipulation on a nano-scale. Robot autonomy encompass the ability of a robot system to operate with a degree of independence from a human operator, and also includes how to distribute workloads between humans and robots. ? Learning robots: A main motivation for introducing robotic learning is to reduce the need for complex and time-consuming reprograming of robots when new tasks need to be carried out (e.g., a need to handle a new part on an assembly line). In NextGenRob, we have developed and demonstrated learning for tracking objects and reliable robot gripping. ? Autonomous operations: We have contributed to developing a structured approach to designing autonomy into current and new applications. A key aspect has been to identify which questions to ask when designing and implementing autonomous system to cover every important area of such systems. ? Human-robot cooperation: NextGenRob technologies and know-how enable robots to cooperate with humans in order to solve task. We have demonstrated how a human can guide an object between positions while two robots take care of supporting the weight of the object. We have also demonstrated how to teach robots new operations simply by hand gestures from the human "robot programmer".

The proposed project will develop fundamentals, competence and leading-edge technologies within robotics to enable the introduction of advanced robotic solutions in specific Norwegian production processes. In particular, the project will target production tasks that today are manually operated, and where robotic solutions are believed to give substantial economic effect and yield a considerable reduction of production costs. As expected, automation of these tasks will eliminate the need for repetitive man ual work improving safety of personnel as well as quality of products and services. It is a key step to increase the international competitiveness of the industrial partners. The project will start out from a set of challenging cases identified by the ind ustrial partners that involve contact/assembly manipulation in uncertain and unstructured environments. By attacking several cases in parallel the project will build the necessary competence and develop generic methods that are relevant for a wide spectru m of similar applications. These tasks are expected to be performed in demanding and uncertain environments, in presence of several another active and moving objects/obstacles in a working area and under the requirement of real time interaction. Such ope rations inherently possess many challenges for developing a system capable to substitute a human. Various innovative contributions in subjects of motion planning, feedback control, signal processing, instrumentation, sensor fusion, robot perception and l earning as well as system integration are to meet in one place. The project addresses the collective technology and competence gaps faced by the industrial partners in these areas. The project will develop new approaches, methods and algorithms for solvi ng the problems and lay down new principles for enabling technologies for automation and robotics-based solutions requested by the Norwegian industry.