The Industrie 4.0 series is written as a collaborative effort from experts in the field. After we at Wevolver wrote an outline of the articles we made a call-out in the Wevolver community for contributors. With their wide range of background and deep expertise these contributors crafted this and other articles in this series and we are grateful for their dedication.
This first article introduces you to the topic and gives an overview of what exactly Industrie 4.0 is composed of.
The term Industrie 4.0 was originally proposed in Germany to rebrand the German manufacturing industry, while truthfully describing the influence of the Internet of Things on industry and the digitization of industrial processes. The terms ‘Industrial Internet of Things’ (IIoT) and Smart Manufacturing are also used to describe roughly the same concept. For simplicity, in this series we’ll use ‘Industrie 4.0.’
In this first article in a series of five, we’ll guide you through the technological developments that make Industrie 4.0 possible. We will give engineering examples by showing you how a modern factory is changing through digitization and talk you through the questions that arise consecutively.
In industrial history, we can separate four revolutions that fundamentally changed the way goods are manufactured. The first started at the end of the 18th century with the introduction of water and steam based mechanical production. In the 20th century, the second phase was defined by mass production and assembly lines using electrical energy, supported by human labor.
After the introduction and later mass implementation of computing power, the third industrial revolution started in the 1970s. Increased control and reliability was enabled through electronics, information technology and automatic production. High reliability was a must, since industrial deployments are often related to critical processes. A malfunctioning could potentially put lives at risk or induce structural damage. To provide such high reliability, industrial deployments during the third industrial era relied on costly and inflexible network infrastructures. This suited the predominant paradigm of linear production: Write highly detailed specifications, have a system integrator implement them, get the line up and running, do some minor adjustments, produce and finally tear everything down and start over for the next product.
Instead of the primary goal being to produce ever larger volumes at decreasing costs, the new challenge is to produce individualized products — or at least an exploding number of variants — at mass production costs. The new need for such customized, but mass produced items, has driven the need to rethink business models, strategy and organizational structures.
This new creed; from highly optimized to highly flexible, asks for a more agile way of manufacturing. The business drivers are changing. From mass production to mass customization. From low-cost country sourcing to proximity sourcing. From distanced automation to human-machine interaction. With the rise of digital enabling technologies and novel manufacturing techniques, industry could evolve. To connect the physical world with the virtual world, the fourth industrial revolution established networked intelligence, integrating the internet of things with the manufacturing process.
From a technological perspective, Industrie 4.0 can be summarized as the trend to incorporate computer aided manufacturing with automation, wireless networks, continuous data gathering, and artificial intelligence. A shift in paradigm. From incremental improvements on existing systems, mechanics, electronics and low-level control, to innovation in algorithms, data, connectivity and usability. Digitally enabling technologies like big data, AI, and 5G are exponentially growing and thus driving Industrie 4.0.
As physical processes are digitized (i.e. represented and controlled in the cyber world), data becomes more and more important. Concepts like ‘digital twin’ (creating a digital replica of a physical entity, like a robotic system) are used to optimize cyber-physical systems (embedded systems integrating computation, communication and physical processes). They can range in complexity, from a single microcontroller chip to complex, multipart devices and are enabling digital representation and control. Rather than general purpose devices, they are often built for a specific task, with relatively lower computational power and low power wireless communication.
As the systems range in complexity and size, so do the manufacturers. Whereas the latest, most intricate implementations of Industry 4.0 might only be feasibly implemented by large enterprise scale manufacturers, small and medium scale manufacturers might have even more to gain. New capabilities can either be production steps that could only be performed manually before — like tasks requiring sensitive force control - or production steps that could not be done at all, e.g. certain parts produced by means of computer aided drawing and additive manufacturing. However, they might be as simple as converting paper-based processes to digital, pulling more sensor data from machines, and running basic analytics on cloud stored data. Industry 4.0 is poised to affect the manufacturing industry across the board.
To show you the influence of these technologies on industry, we will guide you through an imaginary smart factory, Wecycle. This exemplary factory builds custom bicycles from raw material to finished product. On a guided tour we will start by showing how advances in big data analytics, AI and computing enable digitization, then by looking how massively increased connectivity and sensing make communication possible and last by describing the influence of digital manufacturing, digital twins and human robot collaboration.
Through the widespread deployment of sensors and smart devices in current factories, massive amounts of data are gathered. These datasets are referred to as big data. Big data is characterized across four properties: volume, velocity, variety and value. Volume represents the generation and storage of large amounts of data, velocity refers to the renewal rate of data points and their timely analysis. Variety indicates the types of structured and unstructured data gathered from different sources. Last, value refers to the hidden information stored in these datasets. To gain value for the end user, the data needs to be converted using analysis into actionable insights that drive business decisions.
In the industry, big data is gathered through sensors and CPS. It is extracted from industrial processes, then stored, processed and analyzed through machine learning algorithms, and at the end of the cycle translated back to the production process.
This is where we will introduce the first step in our Wecycle smart factory, the arrival of our raw material. Import taxes and border closures related to elections, terrorism, disease and other sources of volatility have crystallized thinking about redundancy and predictive analytics in supply chain design. Predictive factors in trade can all influence the time, cost and availability of shipments by sea, rail or air. By analyzing information about shortage of materials, predictive big data analytics have calculated high probabilities for delays for Wecycle. Warned by previously lowered production output after a similar event, our factory has been able to gain access to a different material provider so work can continue as planned.
The heavy raw materials, in our case aluminium sheets, are precisely moved by autonomous ground vehicles, moving in confined spaces, towards a set of furnaces and hydraulic presses to be shaped into bike frames. Using recognition patterns, learned from previous datasets, problematic defects in the can be found using machine learning. The tiniest deviations in the metal sheets are compared with previous materials. By using big data analytics, real-time scanning allows for faster anomaly detection, optimizing the process. Bike frames with visual defects are automatically expelled from the production line and ordered into different sets to be recycled, repurposed or checked again.
The cyber-physical systems at Wecycle are able to acquire real-time data, perform computations and communicate with each other. The systems are responding to intelligence from within the factory and from across the entire supply chain. Now the challenge is to have a network that can handle these new requirements. Increasingly higher loads on the mobile and cloud computing systems that clean, collect and analyze data are prone to overload older computing frameworks and networks.
The sensor systems at Wecycle generate enormous amounts of data. Wecycle’s smart sensor systems are a combination of a sensor, microprocessor and a wireless communication technology. A collection of those is able to convert a wide variety of inputs (e.g. temperature, pressure, humidity, weight, gas displacement, vibrations) into data and transmit it through the network.
Sensor systems are essential parts of smart factories, and in our Wecycle example, vibration analysis is used to detect defects that could lead to material failure. Due to their networking capabilities, the sensors can work together, being placed at multiple positions next to a vibrating plate. The mechanical sensors are connected with optical sensors to cross reference vibrational data with a visual inspection. Although in-line devices for quality control (like cameras) have been around for a long time, gains in analysis speed and the resolution of sensor data now make real-time defect control possible. In Industrie 4.0 less humans are needed to stand by the line and examine products. Error checking was always time consuming and never watertight. Now it can be automated and executed with more precision than humans can achieve.
The data gathered at Wecycle by the smart sensors connected to the frame production process needs to be analyzed. It can be sent towards the cloud to be computed, or it can be processed at the ‘edge’. An overwhelming amount of data can clog both the gateway and cloud. This is one of the main reasons to use a distributed computing architecture that aims to process data streams at their origin.
Most advanced Industrie 4.0 network architectures are based on ‘edge computing,’ rather than sole cloud computing. A first wave of processing and filtering of the incoming data is performed at the place where it is gathered, relieving computing systems and reducing latency. There is no need to query all data from cloud-based data centers. One can directly and quickly retrieve data from one of the mobile edge nodes. With the use of this technique, the data from all bike frames at Wecycle that have passed inspection is further compressed, while defect analysis is forwarded to the cloud to assist further inspection reports.
Wireless communication technology is of key importance to connect digital and physical systems. Significant advances have been made in sensor development to allow low-cost, efficient communication protocols. Currently, the most used protocol for wireless communication is the WirelessHART (Highway Addressable Remote Transducer) protocol. Released in 2007, there are more than 30 million devices connected. However, due to the increasing demand of wireless networking, the demand has arisen for an upgrade. The extent to which cyber-physical systems can transfer and communicate data is significantly increased through 5G networks. Newer networks will have much higher capacity than current LTE or wireless networks and transmission speeds are promised to be a 100 times faster and have low latency — of less than a millisecond. More so, it manages to provide this capacity in a sensor saturated environment (e.g., in a plant with 1000s of devices).
In this way, 5G enables remote HD video surveillance in our Wecycle factory, or real-time monitoring and feedback between our sensors and hydraulics. On our floor, real-time measurements of temperature are communicated to the furnaces and hydraulic presses, to adjust as needed. If the humidity changes, the pressure drops or the temperature rises slightly, all production variables are adjusted accordingly. This ensures minimal change in the technical specifications of the bikes. The local 5G networks in our factory make it possible for intelligent production components to communicate ad-hoc with each other – without having to install fieldbus cables and configure the communication participants. This makes it easier to move and change different components of the manufacturing process. It also increases network reliability, and promises to lower device cost and energy use. 5G allows functions that were previously located at the central control level to be moved to the edge nodes, allowing controller systems to be leaner. Overall speed is increased through pre-processing and security is enhanced through decentralized storage.
Because our components are wireless, and thus flexible to move, a concept like matrix production can be used. Matrix production allows us to produce multiple interchangeable parts on one single system, thus allowing increased type variety, more frequent changes of models, and quantity fluctuations in production. Now, when a new bike model is introduced, there is no need to alter and optimize the entire manufacturing floor. The design of the modular systems and their placement on the floor can simply be altered. Through its agile design, Wecycle is able to offer personal customization at the cost of mass production.
The increased importance of customization and personalization have led to changes in behavior for consumers and producers alike. The concept of computer aided manufacturing has long been a part of the manufacturing process. Further advances in modelling, simulation and computer aided design tools, combined with continued development of additive (3D printing) and subtractive (e.g. CNC machining) manufacturing practices have made it possible to build shapes and products that were previously unfeasible, both physically and economically.
Real-time monitoring and feedback is assisted by the digital twin (also called ‘digital shadow’). First introduced by researcher Michael Grieves back in 2002, ‘digital twin’ stands for the use of digital models of physical objects to simulate the behaviour of a real manufacturing process. It couples a physical process with a digital equivalent for optimization in a virtual setting. Real-world data, gathered from the print and manufacturing process, is transmitted into the modelled system to complete simulations, validate the system, and dynamically adjust it where needed.[8,9]
Customizable parts can now be mass produced on next generation 3D printers. These are capable of performing real-time quality analysis and real-time adjusting by using sensors and computer vision. When the print is faulty, issues are processed to optimize production. Materials are optimized for batch processing, and through compact and modular design, the production space needed for the printers is minimized. Advances in material science have led to printed parts becoming as strong as injection moulded ones for certain applications.
The handlebars from Wecycle are 3D printed in the highest grade polymers, adjusted to arm length and hand size. Customer length and weight is requested upon ordering, and fed into an algorithm to personalize the saddle shape, fabricated in a multi-part printing process with hard and soft materials. Additional, optional electronics, such as remote controlled lights, digital locks, GPS and automatic locks are added based on customer requests.
The robots used in these processes have gained increasingly larger roles. They are becoming more autonomous, more flexible, and more cooperative. To facilitate the next step in human-machine interaction, collaborative robots (cobots) have been adopted in industry. They have a function to play in the field between manual (assisted) labor and fully automated production.
Collaborative robots allow for new opportunities in which the human worker is in the same workspace, with robotics systems assisting with non-ergonomic, repetitive, uncomfortable or even dangerous operations. A cobot can check, optimize, and document the results of its own work while being connected to the cloud. Thanks to integrated sensor and communication systems, cobots can directly collaborate with their human “colleagues,” safely handle sensitive products, and don’t require a protected space. In order to truly work together, they are programmed to ensure that their behaviour can be tuned or altered by operators, and that they’re increasingly aware of humans in situations where man and machine are dependent on one another. This in contrast to ‘dumb’ industrial robots which will continue repeating pre-programmed movements regardless of what’s in their path.
At Wecycle, a cobot applies adhesive to the frame for custom stickers, saving factory space since the cobot does not have to be separated from the human workers aiding in final assembly. Another cobot helps a human worker to adjust screws in hard to reach, unergonomic places.
Such collaborative working environments are further supported by augmented reality (AR) and virtual reality (VR) technology. AR/VR gives humans the ability to display their steps, or ask for virtual support, either from an AI or from remotely working human experts. They can receive immediate visual feedback, reducing the need to remember complex sequences.
At Wecycle, AR goggles point the workers towards the correct size and position of all the screws used in final assembly. The finished bikes are now moved by a cobot on an autonomous platform, ready to be packaged and sent off.
The Wecycle factory tour is over. In a nearly completely automated and autonomously learning scenario, a personalized bicycle has been produced. The process has been governed by data and automation. By using analysis, real-time feedback can be used to continuously optimize and change manufacturing. The cyber-physical systems on a factory floor connect and work with sensors, machines, humans and artificial intelligence. Flexibility and efficiency in the factory has improved quality and efficiency.
The examples mentioned in this article are some of many, and there are thousands of case studies out there. Investments in Industrie 4.0 continue to be made. Modern times call for processes that are able to run without human interference. In case of a crisis, we rely on industrial production to continue. To feed us, to support our infrastructure or to supply means of transport. Manufacturing is no longer bound to a single discipline. It has become a part of society where computer scientists work together with engineers. A place where artificial intelligence meets production line workers. The full value of Industrie 4.0 is likely to be realized in the next decades and we are surely to see better, more adapted, smarter factories in the years ahead.
This article originally appeared on Wevolver.com, as part of the Industrie 4.0 Deep Dive series. It was sponsored by KUKA robotics as part of their commitment to spread knowledge about the technologies that create our factories of the future.
- Lu Y. Industry 4.0: A survey on technologies, applications and open research issues. Journal of Industrial Information Integration. 2017;6: 1–10.
- Wang L, Wang G. Big data in cyber-physical systems, digital manufacturing and industry 4.0. International Journal of Engineering and Manufacturing (IJEM). 2016;6: 1–8.
- Stojanovic L, Dinic M, Stojanovic N, Stojadinovic A. Big-data-driven anomaly detection in industry (4.0): An approach and a case study. 2016 IEEE International Conference on Big Data (Big Data). 2016. pp. 1647–1652.
- Industrie 4.0: matrix production | KUKA AG. In: KUKA AG [Internet]. [cited 17 Mar 2020].
- Grieves M. Origins of the Digital Twin Concept. 2016 [cited 21 Feb 2020]. doi:10.13140/RG.2.2.26367.61609
- Uhlemann TH-J, Lehmann C, Steinhilper R. The Digital Twin: Realizing the Cyber-Physical Production System for Industry 4.0. Procedia CIRP. 2017;61: 335–340.
- Qi Q, Tao F. Digital Twin and Big Data Towards Smart Manufacturing and Industry 4.0: 360 Degree Comparison. IEEE Access. 2018;6: 3585–3593.
- Origin – Manufacture the impossible. [cited 17 Mar 2020].
- Bahrin MAK, Othman MF, Azli NHN, Talib MF. Industry 4.0: A review on industrial automation and robotic. 2016;78.
- Vysocky A, Novak P. HUMAN – ROBOT COLLABORATION IN INDUSTRY. MM SJ. 2016;2016: 903–906.
- El Zaatari S, Marei M, Li W, Usman Z. Cobot programming for collaborative industrial tasks: An overview. Rob Auton Syst. 2019;116: 162–180.
- Goetz U. Display, overlay, immerse - How automation is benefiting from augmented and virtual reality – KUKA BLOG. In: KUKA BLOG [Internet]. 9 Aug 2018 [cited 13 Feb 2020].
- Paelke V. Augmented reality in the smart factory: Supporting workers in an industry 4.0. environment. Proceedings of the 2014 IEEE Emerging Technology and Factory Automation (ETFA). 2014. pp. 1–4.