Digital Twin: from Automation to Autonomy
NASA introduced the term “digital twin” in a 2010 technology roadmap describing the tools of space travel. Almost a decade later, “digital twin” has emerged as a key tool. It enables a terrestrial space shot. We are being shown the global Industry 4.0’s evolution of Digital Twin: from automation to autonomy.
We talk about “digital transformation” in the Fourth Industrial Revolution, but “digital” has been around since the Third Revolution. Digital includes the form of rule-based computing that reacted like a right-brain limbic system. What’s different now is digitalization that learns, adapts, and develops insights, allowing industries to sense, analyze, act, and optimize. Digital twin technology is an essential element of this evolution.
What’s a digital twin? There’s no such thing as a digital twin or the digital twin. There are, potentially, billions, but let’s start with NASA’s original definition, which remains remarkably accurate today:
“An ultra-realistic, digital simulation of an object or system that uses the best available physical models. Consider the continuous sensor updates and complete historical data to mirror the life of the real-world twin. The digital twin mitigates damage or degradation by recommending changes in use profiles and boosts success-probability of the mission.”
“Digital Twin” sounds simple enough.
The simple name, digital twin sounds innocuous until one considers the breadth and number of industries that benefit from various uses of digital twin technology:
- Aerospace and Defense
- Automotive and Transport
- Machine/Equipment Manufacturing
- Energy and Utilities
- Financial Services
- Healthcare and Life Sciences
- Consumer Goods
The list is extensive and pretty much includes all industries. Within those industries, digital twin technology is used for — among other things:
- Lifecycle documentation
- 3D representation
- Data modeling
- Model synchronization
- Connected analytics
The power of digital twin tech becomes clear when one considers the staggering number of ways the technology can be used, in a surprising number of places.
There is a power that’s an essential element of Industry 4.0’s necessary evolution from automation to autonomy.
What we talk about when we talk about autonomy in digital twinning.
Digital twin, vital as it is — is just part of the evolution to autonomy. What’s equally important to that evolution: AI and domain expertise.
In the case of autonomous vehicles – cars, trucks, ships, industrial robots, etc. — the real world is a dangerous place to train AI. For example, one may recall the autonomous Uber pedestrian fatality in Tempe, Arizona, in 2018.
For safety as well as efficacy, the place to train AI is in a virtual world enabled by digital twins.
Nvidia recently announced an open-source, cloud-based virtual testing ground for autonomous vehicles. Open source will allow companies to train AI in thousands of virtual vehicles across millions of scenarios. It won’t require “in real life” (IRL) presence and lead to zero body count.
In this futuristic virtual world, domain expertise may sound like 19th-century heavy iron. Is the word “experience” too rusty?
In my view domain expertise is the secret sauce of autonomy.
Successful industrial companies and their smart hardware/software partners have learned what works in all sorts of environments and conditions. Combining the right digital twin technology, ever-smarter and more confidently accurate AI and deep domain expertise accelerates the evolution to autonomy.
Industrial autonomy is not binary, black or white, yes or no. There’s a five-step evolutionary journey:
- Human-led/machine supported
- Machine led/human supported
- Machine led/human governed
- Machine controlled
As an industry or a machine ascends toward Step 5, the number of humans required decreases.
People move from being “in the loop” – triggering automated industrial events, even if in a remote-control room – to being “on the loop.” Teleoperators that oversee increasingly autonomous activities remotely are empowered to countermand or correct them.
I believe autonomy will soon develop a Moore’s Law-type “Autonomy Quotient” (AQ). It will determine how many human teleoperators are required to assure the safety and efficacy of autonomous machines and processes. For example, it could be one human overseeing 10 autonomous vehicles or one human to 100 autonomous processes. The higher the AQ, the greater the autonomy.
The ultimate value of digital twins
Earlier I cited NASA’s original definition of digital twin. My preferred definition sounds simpler; “the digital reflection of a physical asset.” That is until one realizes