Title:

Underground Piezo Pest Detection and Identification

Monitoring underground ecosystems traditionally requires destructive excavation. The team addressed this issue by developing a non-invasive, low-frequency vibration data collection system using passive piezoelectric sensors to monitor underground organisms. The hardware integrates piezos with high-impedance preamplifiers, a WM8960 audio codec, and an ESP32 microcontroller to capture ground vibrations. After data is collected, the resulting signal is processed via Fast Fourier Transform (FFT), spectrograms, global bandpower comparisons, and windowed fractal dimensionality. The results of this project show that the system successfully detected underground organisms, and the signal analysis differentiates worm activity from control conditions using features like short-time 5-20 Hz bandpower and ratio-based burst detection. The novelty of this system lies in the design of a custom probe to physically couple piezos to the soil, and the empirical identification of the movement frequency range of worms. Future work will transition from data collection to classification by training machine learning models, such as Ridge Logistic Regression, on feature sets including waveform shape and spectral entropy. Additionally, future iterations will expand the sensor array to enable theoretical spatial localization, utilizing Time Difference of Arrival (TDoA) equations and multipath wave propagation modeling to calculate organism depth and direction in non-homogeneous soil.

Name

Comment*

Comments are viewable only by submitter

Captcha*