A recent European engineering graduate built a handheld radar that classifies wall materials, targeting asbestos detection in residential buildings across the continent. Using a Texas Instruments IWRL6432 boost board paired with an ESP32 dev kit, he created a FMCW prototype that sweeps frequency and captures echo signatures. The system promises a cheap, on‑site alternative to costly lab analyses, aiming for a price point under $100.

The radar’s DSP chain converts chirp echoes into a per‑range, per‑angle density spectre. After mixing the received signal with the transmitted chirp, a range FFT yields distance bins, then Capon beamforming resolves angle of arrival for each bin, producing a sharp angular spectrum. This tensor feeds a convolutional neural network that learns material permittivity and classifies the surface with over 90% accuracy on test samples.

Mechanical housing required custom antenna modeling because off‑the‑shelf TI antennas are proprietary. The builder used OpenEMS, an open‑source FDTD tool, to run parametric simulations and extract TX‑RX transfer functions via Gaussian pulse probing, reducing simulation time from an hour to minutes. Despite funding gaps, the prototype demonstrates that low‑cost radar can identify hazardous asbestos without invasive sampling, demonstrating a viable path for regulatory compliance.