Edge

In today’s digital world, simulating complex systems is crucial for optimizing performance and reducing time to product. SimPy, a Python library for discrete-event simulation, offers a powerful way to create digital twins of real-world systems. Whether you're managing servers, queues, or—as we’ll explore—elevators, SimPy makes it easy to model and test different scenarios. Ready to see how it works? Let’s find out.

In the forefront of technological advancement, drone swarms emerge as a transformative force in the realms of aerial, underwater, and terrestrial exploration. These sophisticated assemblies, integrating robotics, artificial intelligence, and advanced sensor technologies, demonstrate unparalleled coordination and precision. From capturing panoramic aerial vistas and mapping the ocean floor to navigating challenging landscapes, drone swarms redefine the possibilities of exploration and surveillance. Join us as we delve into this revolutionary technology.

The European Union AI Act is a comprehensive regulatory framework proposed to govern the development and deployment of artificial intelligence within the EU. The goal is to secure a balanced and safe use of Artificial Intelligence technologies, at the same time as innovation and technical development is promoted. We at Combine has a broad knowledge of the act, how to comply and avoid sanctions and penalties. Don’t hesitate to contact us if you want to know more.

Given the current state of the industry-standard simulation tools, it is possible to reduce the development time of new products by 50% or more by employing software-in-the-loop (SIL) technologies in the design process.  That is if you have the right model to simulate. What are the varying levels of complexity in vehicle modeling, which model to use, and how to use them? Let’s find out.

Combine’s internal team has taken on several types of data science projects during the last years. Seeing so many different projects in a short period of time reveals the similarities in the approaches and tools that an engineer will use. This has led to the decision to create our own technology platform, where all these different tools are generalized and maintained, which makes them available for all the different projects. The idea is that this will lead to a quicker start-up of projects, a better collaboration between engineers and projects that are easier to maintain.

Ever wondered about the invisible engines driving our digital world? Free and Open Source Software (FOSS) is part of mostly every major engineering component which powers our life. It is in Android phones as the linux kernel, in your banking apps, in the Operating Systems for every microcontroller in your car, everywhere in the web, where it powers engines, websites-generation toolkits, servers, and touches basially everything you can think of. However, despite its ubiquity, many people still wonder where is it, who owns it, how can I use it, and benefit from it. In this blog post we will try to answer some of those questions, delving into the actors, sinergies, and logics of FOSS.

What is MBD? Is MBD too complex for my problem? Will MBD take too much time and slow down our development process? How does Model-based Testing (MBT) accelerate the testing process? Is it a good idea to consider MBT or simulations test results accurate? Do I need MiL, SiL or HiL testing for my system? This post will cover most of the questions regarding MBD and MBT and its feasibility with some use cases.

3 tips on how to get started with Virtual Commissioning

Control system engineering is fast-growing and has played a significant role in technology development over the last decades. This article reviews the evolution of control theory, methods, challenges, paradigm shifts, and the future of the field.

Autonomous Robots – not so different from autonomous cars according to Tesla. Introducing Optimus – Tesla’s AI bot.

Stefan Ristevski, Control Systems Consultant at Combine, recently presented two papers at AIAA SciTech Forum. His research was on the topics “Scheduling Local Information Exchange in Linear Multiagent Systems Through an Event-Triggering Approach” and “Experimental Results of a Quadrotor UAV with a Model Reference Adaptive Controller in the Presence of Unmodeled Dynamics”.

Many people see machine learning (ML) systems as black boxes. Some believe these black boxes contain pure magic and have the power to solve any problem at hand. Others do not trust them at all, for what cannot be explained cannot be counted on. ML is not all so mysterious, though. Just like any other fields of engineering, ML relies on applied mathematics and applied science, and it uses scientific principles to design, build and test a system so that it becomes reliable and works efficiently in real time. Implementing a simple KNN classifier can be your first step to demystify ML.

This blog post may refer to aeronautical applications but Combine have been implementing a model-based approach in a wide variety of industries since 2005. The time saved in the development process is essential for some industries, while quick testing and better products are the main benefits for others.

How relevant is technical debt? Knight Capital lost 465 million dollars in 45 minutes due to unexpected behavior from obsolete experimental code. Zuckerberg changed Facebook’s internal motto from the catchy "Move fast and break things" to "Move fast with stable infrastructure" in 2014. Also in 2014, Google researchers started to publish a series of articles on the (hidden) technical debt in machine learning (ML) systems. What’s scarier than owning technical debt is unconsciously owning technical debt. If the debt stays unnoticed, it will be compounding in silence and become unbearable eventually. As much of technical debt in ML systems is hidden, you should start knowing your enemy now.

An elective usage of irrigation water is required since the world's water demand is increasing. To develop irrigation controllers pursuing this purpose, reinforcement learning implementation could be an interesting step towards smarter automated controllers. This article investigates if Q-learning can be used in order to minimize irrigation amount.

Codex is the latest extension of OpenAI’s GPT-3 large transformer-based language model, pushing the limits of what artificial intelligence can do in the world of natural language generation. One of our engineers was given access to a limited release beta preview of the Codex API and this article explores some of the interesting things they discovered whilst playing around.

Collaborative robots – also known as cobots – have become more attractive compared to the traditional robots previously employed in industrial settings. The main features that distinguish cobots from industrial robots are their lightweight structure, limited power and force, simple interface, and inherited safety guarantees.

Tesla AI Day live stream on August 20 showcased recent improvements from their machine learning teams. Tesla has made it clear that the company intends to solve the problem of self-driving cars. Although the Tesla Bot caught a lot of attention (and undoubtedly is the most photogenic topic of the day), this post will address the other parts of the presentation.

One big part of industry 4.0 is the introduction to big data. New sensors, actuators, and computational power will be introduced in new plants and existing plants. To minimize the environmental impact and the cost of the transformation, it would be great if the communication between the nodes could be done wireless. This article investigates how packaged loss and transmission delay could be handle in a wireless plant.

Some time ago, we presented a blog post where we discussed an efficient tool that we use internally for big-data projects: the Accelerator. In that post, we focused on the performance perspective, comparing it with Pandas. At that point, the features it was providing meant a big blast in the projects we developed, but it still required a bit of ingeneering effort for setup and configuration. Luckly, the developers of the project have spent a lot of time improving that caveat, providing a smooth installation path and therefore making it available to a much wider audience. In addition, latest version also includes some further improvements, like a new user interface that makes up for a better experience.