Vibroacoustic signals (VA) are tiny, high-frequency sound waves generated when materials undergo stress, change, or interact. These signals are often too faint or high-pitched for humans to hear.
Sources of these signals can be:
Friction and wear - When surfaces rub together
Micro-deformations - Tiny shifts or stresses within a material
Impacts or collisions
Fluid movement: Pulsations or flows
Applying machine learning to VA opens a huge field of possibilities. E.g., in a medical context, it allows us to extract information about tissue and differentiate between tissue characteristics such as texture or anomalies, which is especially interesting for robotics. We can furthermore detect events like collisions or underneath pulsating structures when puncuring.
In industrial processes, it allows us to monitor product quality or machine defects.