- The author built a radar startup to detect asbestos in Europe as a graduation project, addressing a common health hazard with traditional detection methods being costly and slow.
- They used a Texas Instrument IWRL6432 and ESP32 dev boards for prototyping, implementing FMCW radar and Capon beamforming to generate material signatures for classification via a neural network.
- RF antenna design was simulated using OpenEMS to optimize performance, reducing simulation time from 1 hour to 2 minutes by using transfer functions and convolution techniques.
- A neural network classifier was trained on electromagnetic signatures to differentiate materials, with a proof-of-concept achieving accuracy across various material layers.