What project do you wish you had the time, the resources and the knowledge to build?

Good question!

I would say 50/50 with smart sensor processing raw data and computing some analitics (i.e. running data. FFT…) and central computing for more intensive model processing.

What do you think?

That sounds tasty! Do you need an high tech greenhouse for tomatoes?

I wonder if there's a way of making a slag mixture that's not too complicated ... then you can make your own mag-stripe cards at home ... sort of a "low-tech" not-a-floppy-disk that sorta uses similar technology.

Wait ... actually ... I guess you could just unspool cassette tape and stick it onto a card ... *scratch that idea*

But making a floppy disk is definitely not a simple adventure from what I can see. Getting the oxide on there is one thing ... keeping it on there is another. Lots of older floppy disks are suffering sticky-shed and losing their lubricant layers, depositing oxide on the head which then gouges out the remaining oxide from the disk with a awful screeching noise.

- Gough

"...I guess you could just unspool cassette tape and stick it onto a card..."

Time to get your ironing board out...

The Secret Genius of Modern Life: Bank Card
https://www.youtube.com/watch?v=crNiK2AiB6E&t=653s

(skip to 10m50s into the video if the link doesn't take you straight there)

It sounds like a good solution. Smart sensors can certainly take some of the computing load and you need the central AI to see the whole picture.

Cool, but it is starting to require a bit of explanation about what else is connected to this super mux.

Not for the faint of heart...

https://youtu.be/FeKxC8pLTbI

I worked on a couple of projects about 10 years back which used 32x32 and 16x16 matrix routing cores. One was to link multiple venues together and the other was for routing signals through a video wall processor. They provided the audio embedding and de-embedding functionality for external audio DSP.

Inputs and outputs were either connected direct to the core or combined with video extenders (similar to HDBaseT) to allow up to around 100m distance from the core. Fibre cards allowed for greater distance (phase 2), but I never got that far (something to do with budgets).

They were later replaced by AVoverIP solutions which basically allows an existing IP multicast network to do the routing.

I love this question.  One thing that I would like to try but don't have the patience to see it through would be designing and building my own pinball table.  Another thing, this one is a little more far flung, would be to make my own robot companion.  This could take the form of a robot dog or maybe something similar to a Star Wars type droid.  Whatever the case it would have to be more than a gadget and have enough 'intelligence' and charm to provide a meaningful emotional impact.

I love how you kept the old notepads and the idea itself alive for more than a decade now.  Very interesting that the big players are visiting your blog site for 'inspiration.'