Innovations Industry 4.0 in production
Discover our latest innovations on Industry 4.0 in production
Industry 4.0 as a term is now well established. The technology, on the other hand, has not yet been implemented everywhere. Artificial Intelligence, Digital Twins, data, localisation, etc. are just a few key words about manufacturing innovations that we realised this year.
Other themes:
Plug-and-play UWB-localisation
People and objects in an industrial environment are constantly on the move. Knowing their location is becoming increasingly important, but the complexity of indoor localisation systems often makes this difficult to achieve. Traditionally, indoor localisation uses fixed anchor points to measure the positions of people and objects, similar to the function of GPS satellites in an outdoor localisation system. Unfortunately, such a system is difficult to set up and, moreover, difficult to adjust once it is operational.
To solve this problem, we developed an automatic, plug and play Ultra Wide Band (UWB) indoor localisation system. The system works without fixed anchor points, but with randomly placed UWB anchors that automatically calculate their relative positions. When a person or object wears a UWB tag and moves between the anchor points, the system will visualise the location in relation to the anchor points. As the system is fully automatic - you only need to place the UWB anchors in the room - setup is 90% faster than a traditional UWB location system.
Precise object position detection by 2D cameras
Robots are getting smarter as Artificial Intelligence finds its way into industrial applications. But to work optimally, AI algorithms need large training datasets, and generating these datasets is very time-consuming and expensive.
To speed up this process, we have developed an end-to-end artificial intelligence workflow that recognises the object and estimates its position. We start from synthetic data that we extract from the CAD file of an object. With this we create a photorealistic training dataset and then train AI models for these tasks. The outcome of this AI model is a 2D vision system that allows robots to perform different tasks, such as bin picking or assembly, without relying on expensive 3D vision systems. Our workflow allows engineers to set up vision-supported robot applications faster. Moreover, this workflow is much more robust and faster compared to what would be possible when taking real pictures of an object. Instead of pulling thousands of photos and collecting them in training datasets, we now use synthetic data to generate a training dataset. In this way, we can generate a dataset in a few hours, which saves a lot of development time and costs.
Data-efficient AI and digital twin technologies for fault detection and prediction of the remaining useful life of industrial machines
Accurate error detection and prediction of the remaining lifetime of industrial machines and production lines are particularly challenging when little data is available. In many cases, even state-of-the-art solutions do not provide reliable results.
Therefore, we developed algorithms for detecting faults and predicting the useful life of ball bearings using data-efficient Artificial Intelligence and Digital Twin technology. Our algorithms outperform current state-of-the-art benchmark algorithms even when only a limited amount of training data is available. The operator can follow everything via a dashboard on which, in this case, the status of the ball bearings can be monitored. These new techniques result in less downtime for production lines, improved maintenance planning and a reduced risk of escalating damage.
Monitoring the sharpness of drilling tools with a smart tool
Drilling into composites is a challenge. For example, delamination or fragmentation of the material may occur during the machining of composites. These problems are often caused by worn drilling tools. Drills, and other milling tools, wear unpredictably due to the specific structure of composites. This makes it difficult to predict when the drill bit needs to be replaced. A good drilling result is unthinkable without a sharp drilling tool.
To ensure this, we have built an optical measuring instrument that reconstructs the surface of the drilling tool in 3D and then detects wear. Because we measure in a CNC, we can check the drilling tool in a matter of seconds. A visual inspection would take minutes. This is not only much quicker, but also much more cost-efficient as no people have to be involved. It also reduces losses in product quality due to blunt drilling tools.
A fast programming solution for multi-degree of freedom robot optimal control
It is hard to imagine a contemporary production environment without robots. Yet programming robots is a complicated and difficult process, especially in applications such as cutting and bending, welding, polishing and laser cutting. Robotic systems must be programmed with a large freedom of movement. Current methods require a lot of manual adjustment and programming and are often sub-optimal in terms of speed and productivity.
To simplify programming, we developed a toolchain that supports the robot programmer during the different steps of the programming process. By using this toolchain, the calculation of optimal robot motion plans becomes much easier, which greatly reduces deployment time. The toolchain offers programmers great flexibility and efficiency. They only need to enter the dimensions and location of the bins and obstacles. The toolchain then automatically calculates an accident-free route for the robot, without the programmer having to manually enter any waypoints. Moreover, the robot's movements are smoother and faster. As a result, our toolchain is no less than 30% more efficient compared to current benchmarks.
Tool for automatically generating and configuring discrete event simulation models
Discrete event simulation models are not only useful for understanding the structure of a production environment, but also for assessing the impact of specific decisions on overall performance. Yet they are still underused, mostly because the creation of such models requires too much effort.
Therefore, we have developed a plug-in for a discrete event simulation software package, Flexsim. With our plug-in, we can automatically generate a model and simulate scenarios based on 'AutomationML' input files. A user can design and configure a new production environment via the user interface. The tool will then simulate the performance of the production line and display the performance. Our plug-in lowers the barrier for manufacturing companies to use simulation software as decision support. In addition, we reduce the time required to create a simulation of your production environment by more than 50% compared to traditional simulation software that uses CAD models.
Smart Factory 2.0
Industry 4.0 is forcing manufacturing companies to invest more in modern technologies to be able to produce customised products at the price of mass production. Yet many companies lack the right knowledge to be 'on board' with these new standards.
SmartFactory is an ecosystem in which we, together with 9 industrial partners, have built a flexible assembly line. With this ecosystem, we demonstrate the current status of the technology. This will also allow companies to evolve towards Industrie 4.0 worthy production systems. Within this concept, we focus on 4 main pillars of Industry 4.0: Standardisation, Digital Twins, Human-centricity and Internet-of-Things. The SmartFactory shows the assembly process of a torch, with the different benefits of each concept: Increased productivity, increased flexibility, increased quality and increased speed.
An offline digital twin for creating added value from your available data
Companies capture enormous amounts of data during their production process. However, it is difficult to extract added value from this data because in many cases it is stored in different ways and locations. In addition, this process requires an enormous technical insight that exceeds the knowledge of the problem domain. As a result, data scientists currently spend up to 80% of their time retrieving data in the correct format, leaving only 20% of their time for the core of their job, which is to gain meaningful insights that add value to the business.
That's why we developed offline digital twins that contain knowledge graphs. Using these charts, your team can analyse the data quickly and flexibly. For example to analyse correlations between product quality and product process. With this solution, you can create additional added value while requiring less IT skills and not having to worry about where the data is actually stored, in which database or in which storage format. This results in a reduction of your data analysis effort and answers important questions about your available data that would otherwise remain unanswered.