January 2020

n TECHNOLOGY January 2020 www.drivesncontrols.com 24 A GERMAN start-up that Festo bought last year claims to be implementing AI (artificial intelligence) directly on machines and interpreting data there in real time. Resolto Informatik says it already has applications at work in manufacturing environments. Until now, industrial applications of AI have been restricted by problems of latency (reaction times) and by the vast amount of data required when connecting to the cloud. Resolto’s Scraitec software interprets data in the field, close to shopfloor machines. It says that this makes it possible to save energy, shorten cycle times and reduce machine failures and production errors. The software knows the healthy condition of a plant and spots any anomalies, providing real-time analyses of sensor data. It is said to deliver diagnoses quickly and accurately, and to suggest recommended actions. “The platform continuously learns from actual operation, integrating the knowledge of the engineers and the customer’s technical experts – we call it the ‘human-in- the-loop’ principle,” explains Tanja Maaß, Resolto’s founder and managing director. The machine learning and AI functions interpret information predictively to optimise parameters actively, or can send instructions to staff via, for example, their smartphones. One end-user already using the technology is the appliance manufacturer, Miele, which was experiencing fluctuating product quality in its manufacturing processes, but was unable to find the causes. Miele operates complex production lines, where items are manufactured sequentially. It is not enough to look at the individual stations separately. The company’s production managers wanted a system that could detect anomalies in complex manufacturing flows automatically. “Deep learning seemed to be the right approach for this,”Maaß explains. An integrated database was developed that brought together different measurement systems. Additional measuring points were configured to support the database. Resolto’s software was used to model the production lines as an integrated system and, in so doing, boosted Miele’s throughput by 1.5%. Anoher application of the technology was at a car-maker and involved a pneumatic clamping system costing just €100. However, unforeseen production stoppages could cost several hundred thousand euros, so the car- maker wanted an early warning system that would indicate any wear and the slowing down of cycle times. The Scraitec software provided real-time data analytics and a learning system that could be used for the predictive maintenance on any type of clamping system. Resolto uses Festo’s CPX-IOT IoT gateway to monitor machines and plants at the field level and to link them to the cloud or Festo dashboards. A software component called ScraiField runs in a small controller close to the machines. A pre-trained model interprets data streams without needing a connection to a component called ScraiBrain located in the cloud. ScraiBrain has access to a host of preconfigured application models. “AI algorithms can be integrated both in the cloud and directly in the components from Festo,”Maaß explains. Resolto says that Scraitec will help users to optimise utilisation of their plants, as well as cutting their maintenance costs by predicting events and recommending action for known fault patterns. It adds that combining its software with plants and machines will transform them into digital tools and open up new business models for machine and plant manufacturers. They will be able to offer new forms of service that add value by the automated coordination of maintenance teams. www.resolto.com Start-up brings real-time AI to machines on the shopfloor Resolto Informatik’s managing director, Tanja Maaß: the platform continuously learns from actual operation AT THE RECENT SPS exhibition in Germany, Yaskawa announced a machine controller that can handle robots, servodrives, inverter drives and I/O systems. When used with the company’s Sigma-7 servo systems, the MP3300iec RBT “singular controller” can supervise 32 real axes as well as up to eight robots. Application programming of the new controller is done using a single software tool (called MotionWorks IEC) with PLCopen and Yaskawa-specific function blocks, which are available free of charge and are said to simplify system programming. Built-in functions also include visualisation tools, camera systems, conveyor tracking, and function blocks for picking and palletising. The controller can be used to integrate robots into existing architectures without needing a programming device or a proprietary robot programming language. The movement sequences of each robot axis – or of delta or customer-specific kinematics – are calculated in the controller’s firmware, allowing programmers to focus on their applications without having to worry about the underlying technology. Machine that include standard motion control axes, inverter drives or I/O systems can be programmed in a single software environment. The real-time Mechatrolink III Ethernet bus is used to synchronise all of the components. If the performance needs to be enhanced in the future, a larger CPU can be added. Machine-builders that use different architectures – robots, delta kinematics or gantry-systems – but similar functions, can exchange hardware without having to adapt their programs. This will increase flexibility and machine throughput, says Yaskawa. ‘Singular control’ handles robots, servodrives, inverters and I/O systems