Comment on Considering positioning strategies for autonomous mechanum-wheeled robots
Saigonauticon@voltage.vn 1 year agoHaha, I know exactly what you mean – I’m most interested in resource-constrained embedded systems. I like the attiny10 a lot. At work I mostly write Python, but in my own time it’s mostly assembly language. It feels more concrete, every decision matters, and anything that goes wrong is 100% your fault as there are relatively few bugs at that level. It’s a lot of fun. Also the datasheet is very good.
I’m self-taught with all the electronics stuff, I paid for it by teaching a course on whatever thing I did most recently. Then I’d use the proceeds to buy tools and parts for the next big (often dumb) idea. I’d also ask for the software engineering assignments from colleagues in those programs, and complete them in my spare time. It was puzzling to a few people why I would want to do assignments, and indeed some were very boring (oh god Java + Spring framework) but others were quite interesting (formal study of algorithms). Sadly, economic reality kicked in and I had to run a company instead of pursuing my education further (I still try to do one ridiculous engineering thing per year though).
I guess there’s a real risk (…like 100%) that I overestimate the motivation students have – so I think I’m going to take your advice and set the level of abstraction with something API-like to abstract away the low-level components (this is closer to my client’s domain). I’m imagining a robot that acts as a WiFi access point, and having something “like an API” that works over UDP packets that describe high-level functions. Then start with something simple – like a digital map with known starting location, and a small obstacle course that can be completed with simple distance measurement, no point clouds. If that goes well, I can develop towards more complex material – probably not full SLAM, but maybe localization on a pre-mapped surface. I have plenty of my own code as examples of how to do simple UDP communications in Python, I could expand it into a custom library.
Sort of like Logo from 1983, but with a physical robot and sensors. I’m a little to young to have used Logo, but the computer lab in my school was really outdated so I got to try it once :D
rufus@discuss.tchncs.de 1 year ago
Well, there’s also turtles to program in Python (i think) and there is Scratch.
Definitely sounds like it. But a motivated teacher is a very good thing. Maybe you’re able to get that spark across to some of the students.
You can always have some extra assignments ready, just in case someone is curious and wants to do/know more. A room full of studens will have a mixed amount of knowledge, abilities and motivation anyways.
I also ate a few books and datasheets on the Atmel chips in my lifetime. Their design is well-thought-out and probably an excellent subject to learn the concepts about microcontrollers.
As of now I like the ESP32. It is ridiculously overpowered if you’re used to something like the ATtinies or old ATmegas. With (at least) 520kB of RAM, two cores that work at 240MHz (depending on variant) and very nice peripherals. Also WiFi connectivity is really useful. But it definitely adds to the fun if you programmed the more constrained (previous generation of) microcontrollers and know how spoiled you are and can feel like a supervillain wasting hundreds of kilobytes of memory deliberately. Or (ab)use some of the peripherals for things that wouldn’t be possible with the few timers available on the Atmel chips. Or do trigonometry at crazy frequencies for your robots, because you can handle 32bit floating point numbers. But I’d agree, that doesn’t teach you the same things if you can do floating point arithmetics for cheap and don’t know if calculating a square root is an easy or difficult thing to do. The STM chips also have nice peripherals. But I haven’t really fiddled around with those.
Definitely hope you’ll have fun being involved in that STEM program.
Saigonauticon@voltage.vn 1 year ago
Good advice all around! Thanks!
I’ve also messed around with the ESP8266 and various models of ESP32. Their WiFi time-of-flight stuff is interesting. I’ve quite a few projects with both actually! My main complaint is that the GPIO don’t behave nicely (also the esp8266 is a power hog and reboots if you screw up the network stack). They are much slower than I’d expect, and have weird states on boot. It’s not too bad to work around this stuff, but I chose the Pi Pico W so as not to have to explain it.
It still blows me away that I can easily do public-private key encryption on the ESP32. And graphics. At the same time!