I'm sure this can be annoying when people do this, but I can't help myself lol. I wonder if you could operate in a different modality and find discontinuities in material properties rather than use it as a classifier. For some reason skin cancer detection popped into my head, but general purpose inspection/detection cases for any discontinuities might be pretty helpful. Depending on the resolution/size of the field it's inspecting a realtime camera overlay might be interesting for correlation sake.
the FMCW tech makes it impossible to have a resolution inferior to 2.5 cm, (so if two layers are appart, you can't physically tell them appart using physical classical modeling techniques with DSP). However, you can use AI to enhance the performance of the system, and make what you are saying possible. The downside of AI is that you need tons of data, which is expensive to get.
So thankful the author posted this. We often learn more from failure than success. Learning from the failures of others is how we can move forward.
The lessons learned at the bottom of the article are gold.
thank you so much for your feedback, it was hard to admit defeat, but at the end looking back at what I built, the parts where I learnt about RF, and just struggled, refactoring the code for the sim (thank god cc is not good enough to understand real world physics functionning for now) were the most satisfying moments
Very cool! Six years ago I worked on a mmWave (76-81GHz) imaging radar with a Rotman lens Tx and Rx. Designed as a LiDAR replacement, but we could see pipes in walls, or detect concealed weapons at ~1km.
32 port Tx (vertical pancake beams) x 16 port Rx (horizontal pancake), something like 60 by 30 degrees. the entire thing used FPGA transceivers as one-bit DAC/ADC, Complementary Golay Code waveforms with one-bit correlation in the FPGAs (two VCU128s) -- digital logic was essentially the same as a binarized neural network, I squeezed a ton of popcnt performance out of those chips using both DSPs and LUTs
Like if you trained a machine learning algorithm to differentiate 10 samples of asbestos containing material from 10 non-asbestos containing materials I wouldn't believe it would work with all the many kinds of materials you would find out in the field in all the configurations that are out there.
All that talk of how the electromagnetic properties of asbestos-containing materials are different are pretty handwavy and lack a theoretical explanation of where the dividing line between different materials ought to be. Overall it strikes me as the kind of half-baked idea that people suddenly feel empowered to do thanks to AI.
It's a cool technology, but for it to gain commercial interest it needs to solve a problem better than the status quo. What problem is it solving and for who? If I was to buy that mmwave radar device it would probably cost more than the $60 test, and I would want assurances that it is as accurate as existing tests.
overtech for a problem that had a solution (asbestos sensing is pretty painful in Europe), but anyways the market was shrinking, and the TAM was totally not VC backable. Tested it out with : wood, copper, alumnium, paper (the book you saw), stone, PVC, plexiglas and air
I'm sure this can be annoying when people do this, but I can't help myself lol. I wonder if you could operate in a different modality and find discontinuities in material properties rather than use it as a classifier. For some reason skin cancer detection popped into my head, but general purpose inspection/detection cases for any discontinuities might be pretty helpful. Depending on the resolution/size of the field it's inspecting a realtime camera overlay might be interesting for correlation sake.
If you'd like to learn more about the module:
https://www.ti.com/tool/IWRL6432BOOST
Does it also work through other materials. i.e. through a drywall etc.
Like if you trained a machine learning algorithm to differentiate 10 samples of asbestos containing material from 10 non-asbestos containing materials I wouldn't believe it would work with all the many kinds of materials you would find out in the field in all the configurations that are out there.
All that talk of how the electromagnetic properties of asbestos-containing materials are different are pretty handwavy and lack a theoretical explanation of where the dividing line between different materials ought to be. Overall it strikes me as the kind of half-baked idea that people suddenly feel empowered to do thanks to AI.
very cool project