Dr. Kegel, when you think about the situation in sensor technology 20 years ago, what is the first thing that comes to mind? What was the status quo in manufacturing and process technology?

Dr Gunther Kegel: In 1998, we took the first tentative steps towards digitalization: AS-Interface, Profibus, Interbus-S et cetera were the first digital fieldbuses available. There were also the first attempts to establish generally binding device descriptions. Profibus and FoundationFieldbus used so-called profiles and device descriptions - DD for short. The first process asset management tools based on Microsoft technologies conquered the markets and the PC found its way into automation. And: sensor technology became an 'embedded system' as the microcontroller was integrated into almost all sensors.

The development of the field device tool concept - FDT for short - as a manufacturer-independent interface standard also took place during this period - with what consequences for sensor technology?

Dr. Gunther Kegel: FDT was the first technology with which field devices - i.e. sensors and actuators - could be integrated into higher-level control systems via a piece of executable code, i.e. a type of driver, but primarily into process asset management systems. The main aim was to develop complex sensors with a large number of conventionally almost uncontrollable parameters, such as those required in the process industry for measuring pressure, fill level and flow rates. Simple sensors for factory automation, such as proximity switches and light barriers, were initially overlooked by FDT developments.

The first discussions about Ethernet in the factory also took place 20 years ago. When did Ethernet become important for sensor technology?

Dr. Gunther Kegel: Although Ethernet has long since arrived in the factory, relatively few sensors are still connected directly via Ethernet today. The majority of users use so-called I/O modules to connect sensors via conventional binary or analog interfaces and thus link them to the Ethernet. While the 4 to 20 mA interface still predominates in the process industry, the 24 V binary interface dominates in the factory. Users are only slowly recognizing the benefits of a digital interface in the sensor, which provides the setting parameters and diagnostic information in addition to the measured values.

What other developments are shaping sensor technology?

Dr. Gunther Kegel: Robust binary sensors, which only record threshold values and transmit two states, have been joined in recent years by more and more measuring sensors. A whole range of new measurement methods play an important role here: ultrasonic and radar technologies in particular, but laser measurement technology has also become established over the last 20 years. The latter is because inexpensive and powerful laser diodes have become just as available as sensitive photosensitive elements up to CMOS chips with several 100,000 pixels.

"Users are only slowly recognizing the benefits of a digital interface in the sensor, which provides setting parameters and diagnostic information in addition to the measured values," says Dr. Gunther Kegel

© VDE, Frank Rumpenhorst

How is Industry 4.0 changing the requirements for sensor technology?

Dr. Gunther Kegel: The most important requirement is connectivity, i.e. the physical and logical integration of each sensor into the digital industrial networks. Ethernet and I/O-Link are used as the physical layer. The industry is currently developing a new physical two-wire Ethernet layer. This enables field devices in the process industry to exchange IP protocols at up to 100 Mbit/s and at the same time supply the participants with an intrinsically safe power supply - a necessary prerequisite for the successful spread of IP communication at field level. Virtual integration, in turn, includes technologies such as OPC UA and FDI, as well as AutomationML and ecl@ss. The sensor is advancing from a measurement value transmitter to a data source for the Internet of Things - IoT for short - and is described completely digitally by its administration shell. This applies to process sensors as well as factory sensors.

Where does process sensor technology fit into all of this?

Dr. Gunther Kegel: By integrating sensors directly into IP communication, process sensors can now use technologies that are borrowed from the Internet. Complex description standards for device integration are being replaced by embedded web servers, which can be used to operate any field device with a conventional browser. New machine-to-machine protocols such as MQTT are being introduced, and the issue of security is suddenly becoming virulent for IP-enabled sensors. It can be assumed that the sensors of tomorrow - keyword Sensor Technology 4.0 - will become more complex because the integration of multi-sensor or multi-dimensional measuring sensors, for example, will become increasingly simple. All sensors must be 'remote-capable' - in other words, they must also allow software updates via the network. Sensor technology 4.0 will therefore also make it necessary to manage software release statuses in the sensors.

Dr. Kegel, would you venture a forecast for sensor technology in the next ten years?

Dr. Gunther Kegel: The sensor as a data source will become an integral part of Industry 4.0 networks. Manufacturers and users will see sensors as digital devices over their entire life cycle, and their software will have to be integrated into release management in a certain way. Artificial intelligence will help to manage software versions and support version management. Connectivity and lifecycle management will become increasingly important for the complex sensors of the future. At the same time, inexpensive, possibly miniaturized chip sensors will become energy self-sufficient and can, for example, be picked up in large quantities by the process material. Machine learning methods are then used to generate 'swarm intelligence' from these simple mass sensors, which generates a detailed, spatially distributed understanding of the process. Three-dimensional laser sensors, which can measure real distances via the time of flight, are used to create three-dimensional spatial images in real time, which are just as helpful for the factory planning of tomorrow as they are for the navigation of driverless transport vehicles, collaborative robots and autonomous machines. Future sensors will increasingly use artificial intelligence algorithms to improve measured values, calibrate themselves, robustly recognize patterns and objects and automatically integrate themselves into complex communication networks.