US teams intend to use vibrations to measure buried rocks on Mars

Before an unmanned robot can safely deploy the lander or excavate on an unexplored planet, the robot and its astronaut operators must know what lies both on and directly beneath the surface. To solve the problem of determining how deeply a rock is buried, a research team at the mechanical engineering department at Virginia Tech has devised a method that involves a fusion of physical vibrations with artificial intelligence. The research, published in the IEEE Sensors Journal, aims to aid robotic excavators in faster clearance of terrain on planets previously impossible to reach using technology or human intervention.

The engineering challenge at hand seeks to increase both the lifespan of the tool as well as the power efficiency and longevity of a robot on its mission to an unexplored planet. Elevation, roughness, and rock height play a significant role in safe spacecraft landings. If a rock blocks what appears to be an ideal landing zone, a small excavator could move or chip it away. However, if the visible rock is actually the tip of a large, buried boulder, attempting to move it risks catastrophic tool damage and wastes critical energy.

Testing vibration tools for robotic excavators

With prior experience as an engineer at the NASA Langley Research Center, Erik Komendera, an assistant professor of mechanical engineering, developed this new technique alongside Ph.D. candidate Yiyan Ruan.

“For as long as humans have built things on Earth, we have had to deal with large and heavy obstacles that cannot simply be ignored or scooped out of the way without careful planning,” said Komendera. He noted that while a human construction worker can intuitively assess which rocks rest on top of the ground and which extend deep below, robotic teams require empirical, sensor-driven data to make operational decisions.

To provide this data, the researchers assembled a set of diagnostic tools to analyze the dimensions of buried rocks without having to break the soil. The experiment included 18 concrete blocks of varying shapes and sizes, each buried at different depths in sand. In order to gather data, team members struck the exposed portions of rock with an instrumented hammer. A scanning laser Doppler vibrometer was used to track the physical vibrations remotely. This setup captured high-fidelity vibrational data without the need to attach physical sensors to the rock face itself.

Artificial intelligence decodes depth data

Following the experimental strikes, the team fed the vibration data into a neural network.

“Estimating how deep a rock is buried using vibrations can be tricky,” Ruan explained. Environmental factors, such as the stiffness of the soil and the angle of contact, cause variations in vibrational response. By selecting key features from the exposed geometry and the vibrational input, the researchers trained a compact neural network featuring just two hidden layers to estimate burial depth with high accuracy.

Using the neural network, the software connects hidden underground profiles with the visible surface geometry and vibrational data. These predictive models can be applied across robotic excavation and planetary exploration missions. For now, the Virginia Tech team plans to expand its database before field deployments, while exploring Earth-based applications such as analyzing unstable rock formations, mapping underground caverns, and identifying hidden structural hazards during disaster cleanup.

Source: https://interestingengineering.com/ai-robotics/us-teams-vibrations-buried-rocks-mars