Digital twins, a concept heavily promoted by NVIDIA on their Omniverse platform, isn’t fully mature. You can create a 3D virtual twin of something, but the twin will be static and physically synchronized with the actual object it is emulating. However, the world isn’t static, and the digital twin, to be fully capable, needs to be tightly tied through sensors to its physical twin to continue to reflect it accurately. It is the difference between a mirror and a photograph, and just like you wouldn’t use a photograph to comb your hair or put on makeup, you wouldn’t use a static digital twin operationally.
To be viable digital twins will need to be increasingly tied to their source material so that what happens to one in real time happens to the other. This goal will require a massive number of sensors to keep the virtual and accurate words synchronized. Not only will identifying and networking those sensors into a control matrix be time consuming initially, maintaining the result over time, if done manually, would be labor intensive.
I recently discovered a product from a small startup, Mapped Inc., that could help solve this problem.
About Mapped Inc.
Mapped was conceived by ex-Cisco VP Shaun Cooley, who helped build the Kinetic Platform at Cisco after observing many IoT failures. These failures were too often connected to an inability to link existing sensors and devices into the IoT framework conceived by IT but provided by the IoT solution provider. Ninety percent of the time and effort in these failed deployments often was connected to the integration process and had nothing to do with the solution provider.
Mapped was created to solve this problem. It uses introspection, pattern recognition, and machine learning to analyze the environment and automatically determine what the sensors and devices do and how they interact. This forms a necessary framework for the creation of an automated command and control structure that is tied to this framework through a normalized API tied to the result. This API cuts across the diversity in hardware to provide an easy-to-use programming interface. It is similar to Enterprise Asset Management in use but moves the model from metadata to actual data.
While I expect this capability to eventually be more closely tied to a more prominent solution provider given this is a small company solving a critical problem and thus an acquisition target, for now, it provides (on paper) a powerful way to relatively quickly and inexpensively develop automated systems managing complex environments. Much of the work can be completed automatically in 12 to 24 hours, depending on the size of the effort.
Currently, the solution is being tested or deployed in grocery stores, microprocessor companies, and primarily managed office complexes mostly to manage lighting, HVAC, Elevators, and large refrigeration units (for grocery stores).
This solution layered on top of a digital twins effort should allow the digital twin to use the connected sensors to update itself to match its physical twin automatically, providing a more accurate real-time representation of the item being digitally duplicated.
Synchronized Digital Twins for Automated Command and Control
While an initial digital twin that closely represents its physical counterpart is possible without Mapped, cost-effectively keeping the two twins identical throughout their like cycles appears to require this Mapped capability.
Using the Mapped technology should be far easier and cheaper to create synchronized digital twins that actively use the large amount of sensors on the site to assure physical changes to the real-world twin are reflected on the digital twin in real time. The technology should also improve the ability for an AI to control connected company assets better and step up to be a viable enhancement or replacement for human operators.
While Mapped isn’t a complete solution, it significantly enhances and accelerates IoT solutions designed to operate autonomously. It may be the missing link in getting to the true benefit of digital twins.