AI Method Estimates Buried Rock Depth From Vibrations

Virginia Tech researchers used vibration data and a neural network to estimate how deep partially embedded rocks extend below the surface.

Updated on Nov. 9, 2026 in Robotics

Isometric editorial illustration of concrete blocks partly sunk in pale sand, with a ripple radiating from one block and a dark sensor arm angled toward it. — Virginia Tech researchers Yiyan Ruan and Erik Komendera developed an AI method that estimates how deep partially buried rocks extend beneath the surface using vibration data. AI Illustration. Upload story photo >

Researchers at Virginia Tech developed an AI-powered method that estimates the burial depth of partially embedded rocks by analyzing surface vibrations. The study was led by Yiyan Ruan and Erik Komendera and published in IEEE Sensors Journal.

Why it matters

Estimating how deep rocks are buried is a key capability for energy-limited robotic excavation, where machines must judge what lies beneath the surface without digging. The approach could support future robotic exploration work in Virginia and beyond.

The team struck exposed portions of 18 concrete blocks with an instrumented hammer and measured the resulting vibrations remotely with a Laser Doppler Vibrometer. The vibration results were fed into a Multilayer Perceptron neural network that mapped 11 measurable features, though the stub does not report depth-estimate accuracy.

The players

Virginia Tech

A public research university in Blacksburg, Virginia, where the study was conducted.

Yiyan Ruan

A Virginia Tech researcher who led the study on estimating burial depth of embedded rocks.

Erik Komendera

A Virginia Tech researcher who co-led the study on estimating burial depth of embedded rocks.

The details

The researchers buried 18 concrete blocks of different shapes and sizes in sand at different depths, then struck the exposed portions with an instrumented hammer. A Laser Doppler Vibrometer measured the vibrations remotely, and the readings were input into a Multilayer Perceptron neural network to estimate burial depth.

November 9, 2026: Article published about the Virginia Tech burial-depth study.

Deeper Dive

The study extends Virginia Tech's sensing-based robotics research by applying vibration analysis to subsurface estimation for robotic excavation. The team plans to keep building datasets and testing the approach for applications on Earth before considering uses such as future rover work on Mars.

For most readers the study is a research development rather than a product they can use now. Its practical relevance lies in future robotic excavation tools that could operate with less energy by estimating buried objects before digging.

The takeaway

Machines that can sense what lies beneath the surface without digging may save energy in future excavation work. Readers interested in robotics can watch how sensing methods like this move from lab tests to field use.