CybEng

the Kalman Filter estimates the true state of a system from noisy measurements. it works by combining two things: • a model of how the system evolves • imperfect sensor readings it runs in a loop with two steps. 1. predict use the system model to estimate the next state. this gives a predicted state and uncertainty. 2. update take a measurement. compare it to the prediction. correct the estimate. the correction is weighted by the Kalman Gain: • trust the model more → small correction • trust the sensor more → large correction uncertainty is tracked using the Covariance Matrix. over time, the estimate converges. noise gets filtered out. the true state emerges. used in: • robotics localization • self-driving cars • drones and flight control • GPS tracking • sensor fusion it is a recursive system that continuously refines its belief about reality under uncertainty.

ESP32-S3-based antenna rotator for satellite dishes cnx-software.com/2026/03/13/dis… KrakenRF Discovery Drive is a low-cost, plug-and-play, fully weatherproof, automatic azimuth/elevation (Az/El) antenna rotator that works with the company's Discovery Dish and other directional antennas, such as Yagis and Wi-Fi grids, weighing up to 5kg. It's compatible with SDR and HAM radio ecosystems, and supports the rotctl (Hamlib) protocol over TCP/IP and the EasyComm II protocol over serial. The C++ firmware for the ESP32-S3 control board is open-source.

Well, it turns out you can just vibe code an multi-target phased array FMCW radar with MTI, CFAR, Kalman filters, and other stuff I don't know, in just a few hours. Once the snow melts, time to take it outside! Running on a CN0566 with a HMC451LP3 power amplifier.

This is (part of the reason) RF engineering is scary

Can you believe this was in a stack of books at work to THROW AWAY? The man himself, Phillip Smith, in hardcover! Snagged it and several other classics: a Sklar and a Roden, too!

The Evangelion anniversary short film got people talking about Eva UI again, so here’s a drone UI I made for a class project inspired by it.

So this is how you can create beautiful curved trace in KiCad

introducing tiny-tpu, a tiny TPU, supporting both training and inference, ENTIRELY on chip! here is an overview of how I got started and how this was built:

@vintagemapstore What happened in 1924 ?

All Locations Mentioned In 200 Years of English-Language Books

Had a fun time building SimSafe for the @nvidiaomniverse Cosmos Cookoff!! AV companies generate millions of synthetic training clips, but bad data with broken physics and unrealistic footage, degrades model performance. SimSafe uses Cosmos Reason 2 to automatically detect physically implausible synthetic AV training data from dashcam footage (by looking at shadow consistency, vehicle dynamics, road texture realism).

We built Talos - a full CNN inference engine running directly on silicon. Every multiply, buffer, and data path lives as real digital logic on the FPGA. This is what deep learning looks like when the model becomes hardware👇

@BToSeeTheLight @iiamkrshn Imagine being an overseas research student, first time ever in UK and your supervisor is a true London EastEnder while the lab manager is a Scottish Highlander.

@iiamkrshn Actually, this happened to me in London once. Guy got on the tube, sat down next to me and starting mumbling something to me in pure Cockney. I was taken aback and asked him to speak up. He repeated his whole long monologue and I still couldn't understand a word, so I apologised.

@tt12514 Υπάρχει και η βιβλιοθήκη 'hackrf_sweeper' η οποία δίνει περισσότερες δυνατότητες από το 'hackrf_sweep' για όσους θέλουν να αναπτύξουν το δικό τους λογισμικό υψηλού ρυθμού σάρωσης. github.com/subreption/hac…

@tt12514 Λογικά το 'spectroscope' πρέπει να είναι ένα GUI για το command line utility 'hackrf_sweep' το οποίο υπάρχει στην βασική σουίτα λογισμικού του HackRF-One και το οποίο δίνει την δυνατότητα σάρωσης φάσματος με ρυθμό 8GHz/s.

Screenshot από το 'spectroscope'. Είναι το αντίστοιχο ρώσικο πρόγραμμα για εποπτεία του φάσματος με φθηνά sdr. Εδώ έχει συνδέσει ένα hack rf και η ταχύτητα σάρωσης είναι περίπου 8GHz ανά sec. Δηλαδή χοντρικά μπορεί και πιάνει πηγές που κάνουν συνεχή εκπομπή.

@tt12514 Ένα άλλο GUI που στηρίζεται στο 'hackrf_sweep' είναι το 'hackrf-spectrum-analyzer' το οποίο επίσης δίνει ρυθμό σάρωσης 8GHz/s. github.com/pavsa/hackrf-s…

We take randomness for granted. Early PRNGs were BAD. Thousands of scientific papers used to rely on RANDU, created by IBM in the 1960s. In 1D space, it looks ok! Map in 3D…you start to see the issues. Now, there *was* a better solution...but it would cost you.

@KaiserMilo3 @lauriewired It was known as SIS back then and indeed the VENONA program was initiated by SIS in 1943.

@lauriewired "Surprisingly, the NSA (likely) knew of these math flaws as early as 1944" That would be very surprising, as the NSA didn't exist until 1952.

I managed to get hold of an HP1345A, the vector display used to film the movie War Games. I've spent the last couple of days interfacing with it and coding for, it, and here's an early demo! Code at github.com/davepl/vector

Your regular friendly reminder: These four datasets all have the same mean, median, and variance. Moral of the story: Always visualize your data! [Link below.]

Using physics to hit a pool shot in the right place