Researchers have developed a robotic navigation system that combines seismic vibrations, magnetic measurements, and touch sensing to map underground environments where GPS and conventional cameras may fail.
Published in Scientific Reports, the study introduces a hybrid simultaneous localization and mapping framework, known as SLAM, integrating multiple sensor measurements through a mathematical optimization process.
The researchers evaluated the system across 23 trajectories covering 10.2 kilometers and approximately 7.5 hours, including simulated settings, controlled underground corridors, and field environments.
The system achieved an average absolute trajectory error of 0.53 meters on the test trajectories, with processing updates taking approximately 57 milliseconds.
These results indicate near-real-time operation under the evaluated conditions, although one loop-detection comparison did not demonstrate a statistically significant improvement over the strongest baseline.
The development builds on efforts to equip robots with multiple senses for navigation.
Such systems could eventually support underground infrastructure inspection and hazardous-environment exploration, but wider deployment will require additional testing under changing conditions.