Edge

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.

We often see the word 'fuel efficiency' being thrown around in our day-to-day life in places like pamphlets, brochures, or even advertisements. Just how important is fuel efficiency, and what pushes the automotive industry to strive towards greater fuel efficiency?

As part of Combine’s continued involvement in Ocean Data Factory Sweden, our team has collaborated with our partners to apply machine learning techniques to the compelling use case of underwater species detection in the Kosterhavet National Park on Sweden’s west coast.

A few years ago, Combine purchased a small submarine. This vehicle is connected to a workstation with a long ethernet cable and is known as a ROV (remotely operated underwater vehicle). During several projects, some degree of autonomous capability has been added to the vehicle, and we are planning to turn it into a fully AUV (autonomous underwater vehicle) in the future. Students from Linköping University have performed most of the developing work and the ROV has been the focus of both master’s theses and project courses. At this moment, we have eight students working on creating a digital twin with an associated simulation environment. The plan is to make the software development as efficient as possible, removing the drawback of needing a pool for every test. The ROV is also an excellent platform to show interested people how to work with model-based development and how to set up HIL (hardware-in-the-loop) testing.

Model predictive control can be an excellent control method finding optimal control inputs to a system. At the heart of the control method is an optimizer. In this blog, you can read about three powerful optimization algorithms and how they perform when they control a vehicle in a complex scenario.

After a turbulent spring, it is finally time to write about some of the new features of Sympathy for Data and the updates in the ecosystem we made earlier in 2020. Earlier this year, we released Sympathy for Data 2.0.0 which was the first version with two different editions, an open-source and an enterprise edition. In this blog post, we will follow up on some of the most exciting changes introduced in version 2.0.0 and the recently released version 2.1.0.

In the last months, Sweden has continued to see an increase in the number of cases and the number of deaths due to COVID-19. One issue that has become apparent in this time is the lack of information from some government agencies in Sweden in other languages than Swedish and English. This puts many in a vulnerable position and at an informational disadvantage that could be costly for everyone in Sweden. A presentation by Folkhälsomyndigheten last week confirmed that the proportion of cases among Sweden’s foreign-born population is much higher than their proportion of Sweden’s population.

The task of calibration is often time consuming, complex, and requires a good understanding of your system and a systematic workflow. We have noticed that some students, unfortunately, are anxious of control systems due to the endless hours spent on calibrating hardware in labs. The main issue often lies in the time it takes to try multiple sets of parameters. To address this problem, we have developed an application to aid calibration on hardware by enabling real-time calibration. In this post, we will describe the concept and how to use it. 

There is currently discussions going on related to Machine Learning and Deep Learning and where this “new” technology will have its use. Does it actually need a super computer to run the algorithms or can you get it running in an embedded system. Since we are a company in the forefront of technology and exist in both Control Theory and Data Science domain we were intrigue by those discussions and wanted to try it out in practice. In this post we will present our finding from our thesis work doing just this!

How do you evaluate your guessing ability when you do not have the correct answer to compare to? The problem is very difficult to answer, so instead of answering it, can we change the problem? Looking at the complex problem of friction estimation for vehicles from a programming and data science approach, can we replace the existing problems with others that might be easier to solve, and if so, how well does that work?

Different games come with different challenges from a decision-making perspective. For this posting we will look at the following illustrative examples: - Chess - Bridge (cardplay only, not bidding) - Poker (Texas Hold’em) - Risk.

So you want to become a Data Scientist? You have followed the online tutorials step-by-step and gotten your first neural networks trained and performing well on the dataset you were given. Now you want to try to do something on your own.. but how? Here we will go through an example process of setting up your first very own deep learning problem and how to solve it. All the way from data collection to final model.

Those who encountered control, communication or other dynamic systems during school time, they dealt most of the time with block diagrams, state machines, electrical circuits or other visual methods to address the design of complex problems. Model-based design methodology will let you easily utilize these experiences when you design embedded software. Besides the simplicity of the design, it has many other assets. However, it is not commonly used in all industries, and in this article, we briefly show why this methodology should be adopted in more industries.

Today when notions such as deep learning, machine learning and even artificial Intelligence (AI) is reaching the mainstream media it is easy to think that an AI revolution is just around the corner. Many people forget that AI have always been “just around the corner” for the last 60 years and the place where we stand today build on many incremental improvements as well as many false starts. It's easy to forget that there already exists many true and proven techniques for image processing that does not require the latest advances in deep learning, and that under the right circumstances can be better to use.

Sometimes you have analog devices that you want to interface to your data processing pipeline but that you cannot retro-fit with a digital interface. Examples can range from reading a pressure valve in an expensive machine to simply trying to tell the time from this old-style alarm clock. Let’s see if we can do this using Sympathy for Data, a web camera, a 3D printer and a few LED’s that was laying around.