Will Artificial Intelligence Give Us the Edge?

The SPS IPC Drives event—held every November in Nuremberg, Germany—disappointed some attendees this past fall by not showing many steps forward for Industrial Internet of Things (IIoT) technologies. Yes, the marketing noise was loud, but little was actually being delivered. I am proud that Hilscher stood out from the crowd with its new portfolio of IIoT products.

With connectivity being our core business, it was clear to us as we initially approached the IIoT concept that some sort of hardware (e.g., gateways) would be needed. After all, IIoT is just another “gateway” challenge, right? Well, perhaps not! During our market research, we realized that we needed to tackle the entire data pathway challenge—from sensor to the cloud and beyond—to get users involved.

As a result, our approach is based on three levels of engagement: sensor connectivity, edge gateways and services—the applications that actually do stuff. We think this approach delivers the right starting point for the automation market.

Despite the many advances around IIoT that have been made in the past few years, as well as the increasing number of real-world applications, I still hear objections to IIoT. For example, I often hear that IIoT obsoletes existing network structures and systems. What nonsense! It should be clear by now that, although IIoT requires new thinking, the technologies on which it is based are well-proven. They are also readily available and easy to deploy. Examples include the lightweight MQTT protocol for transmitting data to the cloud. This has been in use for years in other industries and has just been formally standardized under ISO. Plus, there’s OPC UA, which has been around for a decade.

IIoT actually operates in parallel with existing equipment, taking the data it needs from real-time Ethernet or fieldbus networks without affecting normal operations. Special software is needed to configure what data to choose and where to send it, but that is pretty straightforward.

The reality is that there are increasing opportunities to get your feet wet with IIoT, particularly if you are a machine builder. Of course, you’ll have plenty of questions when you do get started: What does my data mean? How can I use it best? What happens to all that data once it’s been used? Who owns the data? What about its security?

All of these questions bring me to the magic of algorithms (those computer routines that dig out meaning from raw data), Industry 4.0 and the concept of cyber-physical systems (CPS), and artificial intelligence (AI).

The current batch of projects demonstrating AI’s potential shows how far we’ve come. IBM’s Deep Blue overcame the world’s best human players at chess some time back. And IBM’s Watson continues to make headlines in all sorts of industries today, even in automation. More recently, Google’s AlphaGo faced the world’s best Go player, and won! Go apparently has more possible moves than the number of atoms in the universe, so it poses a challenge many orders of magnitude greater than chess. AlphaGo uses “deep learning” technology and relies on neural networks for its computational power. It seems to have taught itself to win, although its builders cannot fully explain how it does this.

The use of such cognitive products and services will be at the heart of IIoT. I saw a great example of this first-hand at Hannover Fair 2015, where Hilscher was part of an extended demo based on IBM Bluemix. In this application, messages such as “OK,” “Warning” and “Error” were sent to a predictive maintenance application in the cloud. Intelligent advice was generated by IBM’s Watson and sent to an operator using services like Text to Speech and Dialog.

We’re still very much in the early days of these technologies, but the odds are good that thinking machines will impact automation sooner rather than later. With IIoT, anything is possible!