MIT researchers in the US have discovered a way to create “digital twins” that can be used to track entire drone fleets. The digital twin is nothing more than a virtual clone capable of predicting situations and suggesting changes to solve problems in real time.
Scientists used a mathematical representation called a probabilistic graphic model to develop the scaled digital clones, making it possible to use this resource in one or more physical individuals, without compromising the autonomy and efficiency of the whole set.
“Custom implementations demonstrated so far typically require a significant amount of resources, which is a barrier to real-world deployment. With our model, we can create digital twins for an entire fleet of aircraft, a wind turbine farm, or a population of cardiac patients,” says MIT engineering student Michael Kapteyn, lead author of the project.
Digital twins have been used in aerospace engineering since the 1970s, when they were used to strategize to bring the Apollo 13 crew safely back to Earth. Computer copying is also part of medicine and urban planning.
Until now, in most cases, each digital twin depended on an individual, customized implementation that worked through a specific application, unable to predict mass situations or reliably coordinate multiple individuals at the same time.
In order to solve this problem, the researchers used a unifying mathematical representation to approximate the relationship between the digital twin and its physical asset. In the case of the drone used during the tests, the parameters of the digital copy were calibrated with data collected from the aircraft itself, making it an accurate reflection of the real drone.
“As the overall drone state changes over time, these changes are observed by the digital twin and used to update its own state to match the physical aircraft. This digital twin can then predict how the drone will behave in the future, using that information to optimally target it,” explains Kapteyn.
The researchers used a 4-meter-long drone to test the new digital twins. During the experiment, the virtual copy was able to analyze data from the sensor installed on the aircraft to extract information about possible damage and suggest changes in its maneuvers to ensure a safe flight.
Digital twin modeling has also proven to be useful in situations where environmental wear and tear can be an important factor in the proper functioning of larger and more complex equipment, such as wind turbines, bridges or nuclear reactors.
The model created by the MIT researchers mathematically defines a pair of physical and digital dynamic systems, linked by bidirectional data flows capable of improving and evolving over time. This makes it possible to develop digital twins for a much larger group of individuals.
“This research can help make the use of digital twins more widespread, since even with the existing limitations, they are able to provide valuable support for decision making in many areas with different applications”, completes the director of the Oden Institute of Computer Engineering and Science at the University of Texas, Karen Willcox, co-author of the project.
The post Digital super twins promise to improve the autonomy of drones flying in flocks appeared first on CmaTrends.