As I previously wrote, lack of  access and reuse to corporate data can be a bottleneck for developing Artificial Intelligence.

One alternative is to re-create the data through advanced simulation. Simulation helps creating massive quantity of data from the so-called “digital twins” of industrial “cyber physical systems”. Obviously simulated data are a simplification so they might miss several real-life issues, but at the same time you are able to experiment with a much wider set of potential situations that are not yet available in real-life – such as a disastrous failure in a chemical plant.

Are simulated data good enough to substitute real sensor-generated data?

 

Advertisements