Working in IoT – smart, connected systems, a testimony

The Flemish research center Flanders Make is nearly bursting at the seams. We’re currently in the process of building a third cocreation facility, focused on Industry 4.0 production. In Kortrijk, engineers and researchers will get to work with new technologies to help companies navigate to the fourth industrial revolution. Our main challenge is to build cooperation between devices and work cells to enable better decision-making.

Smart connected systems

Autonomous vehicles can only see that which is in their own environment, things that are within their ‘line of sight’. For instance, a pedestrian crossing the street around the corner is only visible once the car turns and is in close proximity. I’m currently working on connecting auxiliary data resources to the vehicle. Other passing vehicles or static cameras could issue warnings, but the autonomous car must first be able to understand these. However, if every surrounding sensor starts sounding alarms that something has been detected, it will overload the network and the messages will not come across in time.

To resolve this, I’m working on communication protocols and performing research in wireless infrastructure. You must first examine how the connections will be made. Then, you start testing the characteristics of the wireless communication protocols to get an idea of how you could use them in the best way possible. Often, you discover in the test setup a number of new aspects to consider. For instance, which access points should we provide and what if they’re out of range? There will always be issues, like differences in network configurations – that’s research for you.

In a connected production environment, the algorithms can run in the cloud. The advantage is that we can optimize the production process. We read out sensors and see how it all runs. With algorithms, we can immediately address errors and obstacles in the process. You also want to be able to make predictions. In one of our research projects, for example, we use the strength and frequency of vibrations to predict when a bearing will fail. These same principles also apply to autonomous vehicles. The vehicle predicts the level of risk of the environment on its route, and can then immediately adjust its behavior.

 advertorial 

Device lifecycle management for fleets of IoT devices

Microchip gives insight on device management, what exactly is it, how to implement it and how to roll over the device management during the roll out phase when the products are in the field. Read more. .

Industrial IoT

Technology can only really emerge when it has been validated in an industrial setting. You hear all kinds of promises and predictions on what the IoT can do for the industrial world, but companies are facing a problem: how can they figure this out for their environment? What does the industrial IoT (IIoT) mean for their design and their assembly process?

We’re currently developing the IIoT cloud infrastructure for Flanders Make. My fellow researchers focus on intelligent algorithms, as the models must continuously be tested. In the test phase, devices talk to the cloud. To do so, their set of components (storage, Matlab/Python environment, AI algorithms and so on) must first be configured in the cloud. If we don’t have one overall infrastructure, every project would require a private cloud where each developer would use what they know best and are good at. That would cost a lot in terms of finances and time.

I’m building a platform where you can easily activate these components instead of creating your own implementation. We do this immediately, in an environment that’s representative of the way a company would use it. Setting up your own environment in the cloud would equate to a lot of added work for a business.

This year, we’ll start with four internal labs and two partner labs at the universities, each of which has selected one project. As such, that will allow me to collaborate closely with many researchers. Up to now, researchers worked with local servers located at each lab, only 5 meters away from the equipment. In an industrial setting, however, these servers are usually centralized at a data center – something we imitate to offer an industrially relevant platform.

Data reuse

The cloud has ‘endless’ computing power and storage capacity. On the flip side, however, the communication infrastructure between the cloud and the devices isn’t always equally robust and doesn’t have infinite capacity. That’s why in connected plants, robots and work cells communicate with each other and to the cloud. Therefore, limiting the data streams is necessary due to the inadequate connection to the cloud. With edge computing, we now look to restrict the flow of data and bring cloud functionalities, such as digital twins, closer to the local network.

Hopefully, 5G will improve these communication aspects, with accompanying projects sure to follow. Working in a research environment offers the benefit of being able to switch topics sooner. We can also stop at a lower technology readiness level, compared to industry. If it works from an operational perspective, further development and optimization is up to industrial parties.

Apart from the communication aspects of data, we must also investigate the functional features. In the platform, we add semantics to enable reuse. Ignoring the meaning of data would mean we’d be generating gigabytes and terabytes that are only usable within one project, which isn’t sustainable. If we can maintain an overall structure, however, the data will be reusable for everyone within Flanders Make and all our industrial partners.

Exploring technologies

For me, my work is very interesting because I can explore many new technologies (AI, robotics, digital twins, augmented reality) and learn a lot. In a research center, you’re in the front row of new developments. I’m certainly looking forward to what will come next.