Researchers at Virginia Tech have developed AI-powered tools that use vibrations to determine how deep rocks are buried, which could help robots decide autonomously whether to remove or avoid obstacles. The method, described in a paper published in the IEEE Sensors Journal, could help future robotic missions on Mars avoid wasting energy trying to clear obstacles they cannot move, and could be used on Earth to locate potentially hazardous materials at construction sites and disaster zones.
The study was led by Yiyan Ruan, a Ph.D. candidate at Virginia Tech, and Erik Komendera, an assistant professor of mechanical engineering and a former NASA Langley researcher. Their approach uses a type of neural network called a Multilayer Perceptron (MLP) to interpret vibrational data and identify the relationship between an exposed rock's geometry, the way it vibrates when struck, and the depth at which it is buried.
"Estimating how deep a rock is buried using vibrations can be tricky," said Ruan. "The feedback is affected by the rock's geometry and the way the soil holds it. By carefully selecting features from vibration measurements and exposed geometry, we were able to produce reliable data using a compact neural network with just two hidden layers."
To test their system, the team buried 18 concrete blocks of various sizes and shapes in sand at different depths. They struck the exposed parts with an instrumented hammer and measured the resulting vibrations remotely with a device called a Laser Doppler Vibrometer (LDV). The data was fed into the MLP, which mapped 11 measurable features to estimate the burial depth.
Komendera said that the problem of dealing with large, partially buried obstacles is not new. “Now we have an answer” to how robots can make those judgments, he said.
The team plans to continue building datasets and testing the method for uses on Earth, including analysing construction sites and natural disasters to locate potentially hazardous materials. The tools could also be used to study unstable rock formations and underground caverns and assist with resource and construction planning.
Analysis
Why This Matters
- Enables planetary rovers to autonomously assess whether to dig up or avoid buried rocks, saving limited power and mission time.
- Reduces the need for Earth-based controllers to make slow, uncertain decisions about subsurface obstacles.
- The same technique could improve safety and efficiency in terrestrial construction, mining, and disaster response by revealing buried hazards without excavation.
Background
Knowing what lies beneath a planetary surface is critical for landing and roving on the Moon, Mars, or other bodies. Buried rocks can block excavation of water ice or scientific samples and can damage equipment if a rover tries to drive over them. Current rovers rely heavily on cameras and human operators to spot obstacles, which is slow and cannot see below the surface. Virginia Tech's method gives robots a way to probe subsurface structure using only a hammer tap and a vibration sensor.
Key Perspectives
Yiyan Ruan (Ph.D. candidate, Virginia Tech): emphasised the complexity of the measurement, the feedback is affected by both the rock's shape and how the soil holds it, but said that careful feature selection and a compact neural network produced reliable estimates.
Erik Komendera (assistant professor, Virginia Tech): noted that the problem of dealing with large, heavy obstacles is as old as human construction itself, but that giving robots the tools to judge whether a rock is resting or buried, and whether they can move it, is a new capability.
What to Watch
- Whether the team's dataset expands to include different soil types and rock shapes, which would test the approach's robustness.
- Interest from NASA or other space agencies in licensing the technology for future missions, such as the Artemis programme or Mars sample return.
- Field demonstrations on Earth, at construction or debris sites, that could show the system working in real, unstructured environments.
Checked
11 claims in this story checked against 1 source: 10 supported.
Show each claim
- SupportedVibrations and AI could help robots assess buried rocks autonomously.Universe Today: “These AI-powered tools measure vibrations in surface terrain to determine what lies beneath and how deep structures extend.”
- SupportedThe method uses impact vibrations and a neural network to estimate obstacle depth, with potential planetary and Earth applications.Universe Today: “Their MLP mapped 11 measurable features to identify the relationship between impulse excitation, exposed rock geometry, and burial depth.”
- SupportedThe method could let robots decide autonomously whether to remove or avoid obstacles.Universe Today: “Ruan and Komendera's solution leverages AI to interpret vibrational data, allowing robots to make those decisions autonomously.”
- SupportedThe method could help future Mars missions avoid wasting energy on obstacles they cannot move.Universe Today: “That way, they won't waste time and energy trying to clear obstacles they can't possibly move.”
- SupportedThe study was led by Yiyan Ruan, a Virginia Tech Ph.D. candidate, and Erik Komendera, an assistant professor of mechanical engineering and former NASA Langley researcher.Universe Today: “The study was led by Yiyan Ruan, a Ph.D. candidate at Virginia Tech, and Erik Komendera, an assistant professor of mechanical engineering (and Ruan's advisor) who once worked as a researcher at NASA's Langley Research Center.”
- SupportedRuan said, “Estimating how deep a rock is buried using vibrations can be tricky,” and described the role of rock geometry and soil in the feedback.Universe Today: “"Estimating how deep a rock is buried using vibrations can be tricky," said Ruan. "The feedback is affected by the rock's geometry and the way the soil holds it."”
- SupportedRuan said the team produced reliable data using a compact neural network with two hidden layers.Universe Today: “By carefully selecting features from vibration measurements and exposed geometry, we were able to produce reliable data using a compact neural network with just two hidden layers.”
- SupportedThe team tested the system using 18 concrete blocks of different shapes and sizes, buried at different depths in sand.Universe Today: “To test their approach, their team buried 18 concrete blocks of different shapes and sizes in sand at different depths.”
- SupportedThe team struck the blocks with an instrumented hammer and remotely measured their vibrations with a Laser Doppler Vibrometer.Universe Today: “They then struck the exposed portions of the blocks with an instrumented hammer and remotely measured the vibrations using a device known as a Laser Doppler Vibrometer (LDV).”
- SupportedKomendera said, “Now we have an answer” to how robots can make those judgments.Universe Today: “How can robots make those judgments? Now we have an answer.”
- Partly supportedThe team plans to test Earth applications, including locating potentially hazardous materials at construction sites and disaster zones.Universe Today: “the team plans to keep building datasets and testing their approach for applications on Earth, such as analyzing construction sites and natural disasters to locate where potentially hazardous materials are buried.”