fisian

Posts like these really highlight how much there to think about in robotics. Below the surface of what immediately comes to mind there is just an abundance of topics to address to get an actually well working robot.

Donald L. Unger got an Ig-nobel prize for cracking the knuckles on only one hand for 60 years. n=1 but evidence over a very long time

Agreed. Also being able to use as many aliases as you like (or a catch-all) is very nice.

I started hosting my email at home some months ago and it works great. However, my ISP blocks some ports necessary for running a mailserver (incoming SMTP), so I have to use an SMTP relay (which is on a VPS and does port 25 to some other port that is not blocked). Additionally, to not worry about deliverability, I let my mailserver send all outgoing mail through another service. The additional costs for this are about 2€/month.

I use Stalwart as a mail server and it was not too hard to setup (with some caveats to get it working with my SSL certs behind a reverse proxy. But nothing a bit of LLM chatting couldn't solve). All in all, it's not 100% self hosted, but works well.

However, I don't agree with the suggestion in the article to use a self hosted LLM. My mailserver needs a few hundred MBs of RAM at max, but with the spam filter this would increase like 10x-100x.

There is the --encode (and --decode) options for the `protoc` executable, e.g. `cat myobject.json | protoc --encode MyObject protofile.proto > myobject.bin`

Numberphile has a video with Persi Diaconis about the seven perfect shuffles. It gives a lot of insight and anectdotes in addition to the central theorem: https://youtu.be/AxJubaijQbI?si=ED7ufY4oPZnNxCbd

There's this numberphile video where Persi Diaconis tells a story about bridge and how the players got used to not totally random shuffled decks: https://youtu.be/RIGJH12vVCY?si=cMSx6YEr8MMFgYg6 (Starts at around 3:30).

Discussion at the time: https://news.ycombinator.com/item?id=26591669

I interpreted it another way: the boxes are what connects the past to this written story. Without the boxes most details would be forgotten and this article wouldn't exist. The documents in the boxes were only thoroughly read by one daughter starting in 2024.

I didn't know CMake introduced this debug adapter. This would have been very useful to me at some occasions.

Instead, I used printf debugging (aka `message(FATAL_ERROR )` debugging in CMake).

With this debug adapter, there also seem to be integrations in VSCode and CLion. I will have to try them out soon.

I think this is a great introduction to logical thinking and coding. The overcooked scripting layer looks awesome and very polished. Reminds me a bit of Scratch (the programming language). Are you going to make it available to others?

There are also video games based on this concept, e.g. Bots are Dumb. So maybe your scripting layer it could even become its own commercial game.

For the part which talks about turbulence damping, this page illustrates how this looks like on a smaller plane: https://turbulence-solutions.aero/technology/

(Not affiliated, just saw their demo once.)

This principle is also highly relevant in safety critical systems for using redundant sensors. Just adding a second sensor is often not enough. Because if they disagree, which one do you trust.

One example of this is in airplanes.

Looking forward to the early bird launch. JMAP support from an email provider (other than fastmail) is great. Additionally, this seems like a good way to fund thunderbird apart from donations, while supporting open source email.

Another fun thing are these stable diffusion/controlnet combinations which create QR codes that at the same are AI generated art. e.g. qrdiffusion or qrbtf

Does a Banana Pi BPi-M5 fit your specs? The banana pis have pretty good networking options.

The reported 119GB vs. 128GB according to spec is because 128GB (1e9 bytes) equals 119GiB (2^30 bytes).

off topic: What's the font this website uses for the code? The font ligatures seem nice, but I also would have to get used to reading code like that.

I found this course very helpful if you're interested in a bit of math (but all very well explained): https://diffusion.csail.mit.edu/

It is short, with good lecture notes and has hands on examples that are very approachable (with solutions available if you get stuck).

It's just an ad for their SSH cert service...

I feel like for SSH certs to expand beyond large companies, there's the need for an open-source service which does the issuing of short-lived certs after a user authenticates. I know smallstep, but their offer feels open-core/freemium.

Can anyone recommend a site for buying DRM free books so I can read it with KOReader?

Denying math has a tradition among politicians: https://en.wikiquote.org/wiki/Malcolm_Turnbull

Mathematically, what you need is the binomial coefficient ("n choose k"): https://en.wikipedia.org/wiki/Binomial_coefficient

GrapheneOS on a Pixel is probably the most polished and secure experience. I have installed it (and enabled sandboxed Google services) on my mom's phone (she's pretty non-technical) and she had no bigger problems in the last years.

There are the anti prime numbers (also called highly composite numbers).

Yes, the Lissajous https://en.wikipedia.org/wiki/Lissajous_curve which also can be turned vertical (to look like an 8 instead of ∞).

The official Raspberry Pi site [0] seems to be a much better resource than this article. It also says that the AI kit was surpassed by the AI hat+ [1] (which is a bit cheaper, too).

[0]: https://www.raspberrypi.com/products/ai-kit/ [1]: https://www.raspberrypi.com/products/ai-hat/

I also understood it to mean the "move fast, break things" attitude, that many techy VC startups have.

Great work!

I would've needed this recently for some data analysis, to estimate the mass of an object based on position measurments. I tried calculating the 2nd derivative with a Savitzky-Golay filter, but still had some problems and ended up using a different approach (also using a Kalman filter, but with a physics-based model of my setup).

My main problem was that I had repeated values in my measurements (sensor had a lower, non-integer divisible sampling rate than the acquisition pipeline). This especially made clear that np.gradient wasn't suitable, because it resulted in erratic switches between zero and the calculated derivative. Applying, np.gradient twice made the data look like random noise.

I will try using this library, when I next get the chance.

Also check out the "promo" video for this https://youtu.be/EHqPrHTN1dU