 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : Mites and AI
| Author | Message | ||||
PhilP Newbie  Joined: 25/01/2017 Location: United KingdomPosts: 31 |
There is a lot of discussion about Artificial Intelligence at the moment with a global meeting on the topic taking place at Bletchley Park in the UK. (Incidentally I did my training at Bletchley Park in the early 70s). One aspect of AI is machine learning whereby computer can learn based in their experience automatically. Examples include robots learning where to travel based on their travels around an area. Or perhaps Tamagotchis who can behave a bit like humans. In an idle moment I was wondering what could be done within the Mite family to support AI? Obviously, it would not be possible to produce a ChatGBT but there must be something that could done to demonstrate AI. I think it would have to be a software intense project which might be challenging and may require networking of Mites. I guess the CMM2 is the most powerful? Anyway it might be fun to link Mites with AI. |
||||
Mixtel90 Guru  Joined: 05/10/2019 Location: United KingdomPosts: 6798 |
You could probably do something that *looks* a bit like AI - something a bit more convincing than Eliza nowadays. You need lots of fast-access storage really, I suspect. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 4243 |
I have no clear idea the border is between (artificial) intelligence and learned knowledge. Most what is called AI today er pure "tuned" algorithms. Feed 1 million pictures to a large computer, and it starts recognizing faces. That is not new, only it the scale is different. When I was in my twenties, a large chemical industry already used "self tuning algorithms" to increase the output of chemical processes. I think this is called AI today. In it's simplest form this can be done today with mites easilly. I myself used "self tuning algorithms" for a small robot. I put the robot (2 wheel type) in a box, with end sensors, and made it run overnight tuning it's PID parameters. That was not hard to do. Only took a few nights before it gave good results, since you need to set boundaries within the algorithm is allowd to tune. Volhout PicomiteVGA PETSCII ROBOTS |
||||
karlelch Senior Member  Joined: 30/10/2014 Location: GermanyPosts: 172 |
There are machine learning models that run on microcontrollers. Once example for the Arduino is here. The workflow is typically that you train the modell on a "real PC" or, recently more in the cloud and then transfer the trained model to the microcontroller. A common platform for this is TensorFlow light. But this uses C(++) code for performance reasons. One could imagine two ways of making a PicoMite "more intelligent" with current ML approaches: (1) add a ML processor to its side (e.g., a controller running such a pretrained model) and access it's "intelligence" through some interface (a serial port), or (2) use a WebMite and let it communicate with a model running online in the cloud. Just some thoughts Thomas |
||||
Rickard5 Guru Joined: 31/03/2022 Location: United StatesPosts: 463 |
Phil My entire Life, being a Child of the '70's and '80's, I have been Obsessed with AI and expert systems! Tamagotchi has been a Life Long Dream of mine and now that I've built a few of the lower cost Robot Dog Kits powered by Raspberry Pi, the Tamagotchi idea comes straight to the top. unfortunately my programming skills suck to much to create a Tamagotchi program for them something that has an algorithm that balance attention feeding and actual walking the dogs with health and Happiness and moods. even a sleep play balance and collision avoidance autonomous wandering around the house  and as excited as I am about the emergence of A.I., I fear it with the Lunatic in charge in the Colonies, the future of A.I. is something to FEAR ! it's time to stop the limitations on A.I. don't let the government control innovation I may be Vulgar, but , while I'm poor, I'm Industrious, Honest, and trustworthy! I Know my Place |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 6798 |
It can be difficult to differentiate between AI, a LLM and a sophisticated state machine. There can be overlap but there doesn't have to be. IMHO there isn't any true AI yet as it would imply the ability to think. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Amnesie Guru Joined: 30/06/2020 Location: GermanyPosts: 396 |
Hello, I share the toughts about a Tamagotchi - although this isn't AI - I think it would be possible with MMBASIC and fun, too. But I am neither able to do the programming... Maybe at some point in time. But I really like the idea! Greetings Daniel |
||||
| DaveJacko Regular Member  Joined: 25/07/2019 Location: United KingdomPosts: 76 |
who was it that said, when asked if a machine could think, replied something like "can a submarine swim ?" possibly a Mr Dykstra. do have a look at some of his other quotes quite entertaining and thought provoking |
||||
 Print this page |
