Autonomous production of the future configures and coordinates its production cells itself. Pioneers are already implementing the characteristics of the smart factory. However, humans also remain irreplaceable in the smart factory.
In Sick's sensor factory in Freiburg-Hochdorf, driverless transport systems curve around automated production modules and workstations operated by people or collaborating human-robot teams. “The modules are cells in which the robot performs a defined task in a fixed working environment, such as the final assembly of various sensor components,” Joachim Schultis explained, Head of Operations for Photoelectric Sensors & Fibers at Sick AG “The modules are completely setup-free; format and material changes are carried out by the control system operating in the background.
On the way to the smart factory, Sick has transformed conventional assembly line production into a networked production matrix. Machines, tools and means of transport are equipped with sensors and actuators, i.e., eyes and ears, hands and feet, and networked via the Internet of Things. Industrial analytics evaluates what occurs in process and sensor data in such a smart factory scenario. These include artificial intelligence (AI) methods such as machine learning. The data are used for plant control, provide information on the efficiency of the ongoing processes and information for plant optimization. For example, knowledge of machine conditions enables predictive maintenance.
Sick is not the only smart factory pioneer in industry. Other companies have switched from assembly line production to react to a variety of types, frequent model changes and quantity fluctuations. For example Kuka: At the heart of the matrix production of the plant and machine manufacturer in Augsburg is its Smart Production Control software. It controls AI-based flexible processes, in which Autonomous Guided Vehicles (AGVs) adapt standardized production cells according to orders and supply them with material. Dürr Systems AG even supplies its process – painting systems – as a modular concept. The linear system configuration is broken down into boxes, so that automobile manufacturers can guide variable quantities and car body variants through it. In addition, orders for new models are easier to integrate and capacity is easier to scale.
For Sick, the investments in its smart 4.0 Now Factory have definitely paid off. “Two new product families have been added to the initial five product families produced there. In addition, the number of industrialized variants within the product family could be increased significantly,” Sick Manager Joachim Schultis stated about the first advantage of the investment. The smart factory can also flexibly meet customer requirements for small quantities, which are becoming increasingly frequent due to the trend towards individualization in the industry. In addition, the control software prioritizes orders, production steps are optimized for each product, automated quality control identifies faulty modules and the cause of product defects is shown via the production history.
Production technology and AI expert Professor Martin Ruskowski, Head of the Smart Factory KL Technology Initiative, believes that the vision of a fully automated, human-empty smart factory, in which machines control and optimize themselves, is a “faulty concept”. Thanks to automation technology, machines can often take on repetitive activities, and AI technologies can handle demanding work with data, for example, evaluate large amounts of data. “But AI has nothing to do with real intelligence and creativity. It will never suggest a creative solution to problems. People are irreplaceable for this task,” Mr. Ruskowski emphasized.
Therefore, the smart factory is primarily about introducing assistance systems to support automation and production systems, because this pays off quickly. In sheet metal processing, for example, conversion, assembly and process planning require labor-intensive manual work. Four hours of indirect work can quickly be carried out in one hour of processing time. Materials and tools must be sought, there are long distances involved in preparing materials, and vehicles must be available for intralogistics.
Ultimately, the way to the autonomous factory is comparable to the way to autonomous driving. It will take some time before cars can drive safely even in busy city centers on an autonomous basis without drivers. First, assistance systems are installed in vehicles (automatic distance control, lane departure warning) and then the development stages of semi-automated, highly automated and fully automated driving are required until the final stage of autonomous driving is reached at some point.
Every company has to define for itself which autonomy level is technically and financially beneficial for production, i.e., its way to a smart factory: This concerns the construction of a new factory, the integration of advanced systems into existing production or process optimization. “Anyone who wants to go new digital ways in production and implement industry 4.0 principles does not have to build a new production plant,” Joachim Schultis said, referring to Sick's sensor-based solutions, which make data from sensors visible and generate added value for an existing factory. “This enables companies to collect statistical data, analyze it and take the step into Industry 4.0.
Individual measures of smart system optimization also make the automation beginner a high achiever. For example, the analysis tool "Darwin" from Plus10 GmbH, a start-up of the Fraunhofer Institute IPA, finds defects in fast-moving production plants via self-learning algorithms and performs automated machine benchmarking. Using it, automotive and pharmaceutical companies have reduced cycle times by up to 18 percent.
But what are the stages on the way to autonomous production? Researchers from the Scientific Society for Production Technology (WGP) have divided the transition from human to machine tasks into six stages in the study “Industrial Workplace 2025” (see box). The higher the degree of automation, the more the production plant autonomously makes adjustments to compensate for disturbance variables such as machine inaccuracies, fluctuations in material properties, machine wear or tool wear.
Based on the stage model of automated driving, the Scientific Society for Production Technology (WGP) has developed a stage model for automation in industrial production. This makes it possible to assess the current state of development of a company on the way to autonomous production
1. Stage 0 (operator only): The operating personnel adapt the production system to all disturbance influences that occur.
2. Stage 1 (assisted): Inaccuracies in the production system are reduced by controlled drives. Machine operators adapt the production parameters to the product specifications.
3. Stage 2 (basically automated): The production system regulates individual, selected process parameters according to the specifications. Machine operators adapt the production parameters to the product specifications.
4. Stage 3 (basically automated): The production system automatically regulates individual, selected product characteristics. Adjustments to changed production conditions are made autonomously for this purpose.
5. Stage 4 (highly automated): The production system automatically regulates all relevant product characteristics. It can eliminate all deviations from defined properties and monitor its own system limits.
6. Stage 5 (completely automated): The production system regulates all relevant product properties, recognizes and eliminates explicitly and implicitly specified fault patterns. It recognizes and expands its own limits.
By the way, the Industry 4.0 maturity index of FIR e. V. at RWTH Aachen also goes over six stages, which describes the development path toward the smart factory. The stages have the following designation or meaning there:
1. Computerization of production
2. Connectivity (networking of production facilities)
3. Visibility (“seeing through the use of sensors”): What happens in my production?
4. Transparency (“understanding thanks to mass data analysis”): Why does this happen?
5. Predictability (“prepared thanks to predictability): What will happen?
6. Adaptability (self-optimizing thanks to automatic action and self-optimization): How can we react autonomously?
Stage four of the WGP model reveals where the limit to practical automation currently is. This is also the subject of current research on the smart factory. The production system automatically regulates all relevant product characteristics, eliminates deviations and monitors its own system limits. Nevertheless, people remain indispensable in autonomous production: They monitor all other product properties and intervene to make corrections if system capabilities are exceeded.