karmicthreat

I will always remember him for his role in gun men

This looks like ROS2 for missiles.

There are already some products for arms that do this.

I had better luck recently. I’ve tried most of the commercial systems. None of them really got very far into the design.

Recently Radu over at Galvano.ai let me beta his platform. Way farther and way faster than anything I have used so far. Even managed to make the schematic placement sane.

The actual project was a regen clamp. So if the output voltage is greater than the input voltage (back emf from motor) you divert the excess to a resistor.

This has some analog tuning needed. So I created a python and skidl mirror of the netlist. This mirror was used to create ngspice unit tests to drive component optimization. Turns out not every resistor needed to be .1% and the ESR of our storage caps was a critical value.

All in all, it’s like ai and software. It’s not going to replace every job but will change how the jobs are done.

What’s currently the best pattern if I want to combine Fable and GPT if a workflow but keep using subsidized tokens?

I used company money on it. I was hoping I could massage it along enough to get a workable test fixture out of it. I wanted to put together a simple hardware-in-the-loop tester for a component of our product.

Adafruit probably did a review of AI PCB tools. I've used Flux.ai before; it was a pretty bad experience. After about 50-100$ in tokens a couple of times, I couldn't get more than a couple of simple components on the schematic. And not in sensible positions.

The product just grinds tokens for little return, in my opinion. I had far better luck wiring together KiCad MCP, SKIDL. There are some AI-driven autorouters out there now. Placement is probably the big issue that needs to be solved now. I could only get about 80% of what I wanted together with my hacky workflow.

If you want something more industrial Chinese arms that ape UR are 10-20k.

I'm tired. I've been at it since '89' when I was 15. Everything being on fire all the time (in the world and in my current robot project) is just burning me out.

Most of my curiosity is tempered by how it can make me money.

I do appreciate that 50-70% of the boring work can be done with AI agents now. As long as you know enough to have opinions and guide the process along, it can be helpful.

Expertise and learning don't seem to be AS important with the upcoming gen of developers. However, there is also so much out there now that it would be much harder to start from zero as opposed to being there from the beginning.

But I think gen X and Millenials were probably peak interest and curiosity, now it's just a job for the later generations.

Is there a way to translate these into friendlier names? It would be nice if something like lidar-front.robot1.yada-fleet.bar could be made.

I’ve been doing a lot of work with ros2 lately at work. Been seeing some moves with VLA models driving robots with text prompts. I think I’m going to focus more of my “free” time on mastering that technology.

This seems kind of like teleport without the logging and remote infrastructure.

I've been "Vibe Coding" this weekend with the vibe of a micromanaging tech lead or PM. Actually, it's not terrible if you just accept that you need to treat the AI like a year 0 engineer who is really good at googling.

I've run the project through cline and roo. Also tried Claude 3.7 and the 1M context Gemni 2.5 pro model. I'd say Gemini is less creative. But it's still good.

I can see how it is a productivity booster. Or at least it gives you the illusion of one. Really I think best part of it is just building out detailed documentation. That's really the killer app for me.

It's fairly nice just to build a simple embedded UI. There is also a useful editor now, though I haven't used it. It was implemented as a stand-alone program, I would have preferred a VSCode plugin.

It can be a bit difficult to get going on a specific platform, but once you have it going, it's smooth sailing.