How often do you hear about a “leapfrog” jump in technology? Every month? Every week?
Before, a true leapfrog was treated with reverence and the term was used sparingly — like when developing countries skipped the use of phone landlines altogether in favor of mobile telephones. But recently, the term has lost its meaning. Every new tech craze seems be labeled a leapfrog.
But today we are witnessing advancements in technology that will restore the word “leapfrog” to its former glory— technologies which give machines the intelligence and power to understand and control their own behavior.
This advance is enabled by two key components. The first is microsystem technology with artificial intelligence and machine learning. These microsystems put the “sense” in modern sensors, allowing drones to self-navigate and cars to self-stabilize. The second is comprised of the controllers at “the Edge” of the Industrial Internet that lets machine-learning go beyond the cloud.
Consider microsystems. They are smaller, require less power and are now pervasive across industrial systems and machines. What could only be accomplished by a series of complex devices before is now done by one: embedded computing systems called microcontrollers. Thanks to increased processing power and lower costs, assets can now control aspects of their own behavior without requiring additional tools or information from outside. This is beginning to happen in the automotive industry, where microcontrollers are allowing cars to drive themselves.
And it’s happening at the “Edge” of computing, where devices become their own small data centers. New controllers operating independently will be able to self-monitor, communicate with others in the network, and adapt their operation based on the data they collect. They will harbor ‘digital twins’ of themselves to predict their own health using AI and machine learning.
GE is beginning to exploit these capabilities with drones to inspect flare stacks at oil and gas production sites. If wind gusts suddenly pick up during a site inspection, the drone can – on its own – adjust its position to ensure it remains a safe distance from the flare stack to avoid colliding with it. At the same time, the drone also knows to adjust its camera to continue to capture the clearest possible images.
The Industrial Internet is where machines learn from an overall network. The Edge, on the other hand, is the next level, where machines learn from their own experience and integrate learning developed in the Cloud. It’s just like how people learn in a classroom versus learning from the trials and errors of the real world. The former will only get you so far; the latter lets you leap forward, building on classroom knowledge with real-time intelligence and individual experience.
Machines are smarter at the Edge: more efficient, more powerful, and less prone to breakdown and maintenance. Using the same drone inspection example, imagine drones performing inspections of hundreds of flare stacks at oil and gas production sites all over the world that get pooled and analyzed in the Cloud. Through these collective experiences at the Edge, a best practice could emerge that identifies the optimal frequency of inspections that provides the earliest detection at the right cost point. This best practice can then be sent back down to individual drones at the Edge to optimize their inspection procedures. Here lessons from the class inform each student to make them smarter.
As Edge Computing takes hold, this leapfrog moment for machines has vast implications for industry. It will allow corporations to maximize equipment performance and uptime, extract maximum efficiency at limited costs and also limit environmental impact. This is a leapfrog truly worthy of the term.
(Top image: Courtesy Getty Images.)
Colin Parris is the Vice President for Software Research at GE Global Research.
Danielle Merfeld is the Vice President & General Manager of the Niskayuna Technology Center, GE Global Research.
All views expressed are those of the authors.