Industry 4.0 refers to the monitoring, storing, analyzing and acting on data gathered in a manufacturing environment. We know this. Today this is typically implemented in partnership with one of the incumbent IoT platforms from companies such as Siemens or GE (Perdix).
These vertically integrated systems offer hardware, software and maintenance from the same vendor.
Sound familiar? It should. There are clear parallels between these massive systems and the mainframe-computer manufacturers of the 80’s.
These systems work. They are reliable. But they are complex, expensive and closed, making it difficult to mix equipment from different vendors. However, disruptive forces are on the horizon and already getting deployed by some of the more adventurous and innovative manufacturers.
Let’s take a look at some of the disruptive technologies enabling this:
Public cloud computing—AWS, Google and Azure are all computing giants offering global, state-of-the-art data facilities that can be accessed through any browser. This means investment in IT infrastructure is driven down, data can be aggregated from geographically diverse locations and customers can choose pay-as-you-go models.
Low cost AI running at the edge—Nvidia, Intel and Google all offer AI chips that, for around $70, can execute machine-learning algorithms. This opens up applications like object-recognition and machine-vision, which could not have been deployed in the past without a huge expenditure.
Single-board computers—Initially aimed at the educational market, these products are now finding their way into the enterprise. The market leader is Raspberry Pi, but Arduino and Beagle Bone also have market share. These companies offer a $35-40 product and are on the precipice of replacing expensive IoT-gateway hardware that is currently in use.
Low-cost cameras—A 4K camera costs around $30 and, when combined with a single-board computer and AI chips, enables a complete object-recognition system. Multiple cameras can be deployed to perform 3D-positional tracking (with some triangulation calculations), enabling tracking and real-time analysis of goods moving through your facility. These systems also help identify bottlenecks and enable future-layout planning.
Digital twinning—Displaying all this data on a scaled and accurate 3D model opens up all sorts of visualization and analytical approaches. Digital twinning is ideal for displaying large amounts of incoming sensor data, as opposed to wading through endless dashboards. The cost of creating the digital models is rapidly dropping, making this a useful approach even for smaller plants.
Seamless edge / cloud computing—IoT systems often need to run data-processing at both the edge and in the cloud, depending on the type of data being processed and latency requirements in the system. This area is now receiving a lot of attention and solutions are starting to appear that make moving algorithms between edge and cloud seamless for the user, which removes complexity from the system and reduces implementation and operational costs.
Putting all of these elements together creates a functional IoT system that can be deployed economically, even in a larger facility. The hardware components used are inexpensive—replacement, rather than repair, becomes the default option. There is no need to re-work a $36 gateway computer…simply swap it out for a new one.
Returning to the mainframe analogy, today we are seeing low-cost, off-the shelf hardware networked together—a bit like when network PCs started entering the workplace, which had a significant impact on mainframe-computer manufacturers.
Expect to see some of the same forces in Industry 4.0.
John Burton is CEO of UrsaLeo.