One of our guys has been talking about these, but hasn't gotten one yet. Using one for docent duty is a fantastic idea, that is seriously cool and sounds like it'd be a neat learning project, whether it works out in the end or not. -Dave On 12/22/25 12:15, Bob Roswell via vcf-midatlantic wrote:
I got an early holiday present. The Hugging Face/Pollen Robotics AI robot. https://www.pollen-robotics.com/reachy-mini/ The robot is open source, and all the programming is in Python. I built it over the weekend, and it has a reasonable chatbot built in. As a fun exhibit, I would like it to be an assistant docent in one room of our computer museum.
As is, I can ask it about many of the artifacts in the museum, and it already can give a one sentence summary. But it is not yet up to the task.
1) I can get more details, but it takes multiple prompts. I would like a 2-3 paragraph answers. 2) I can "train" it to learn additional facts, but the training is not persistent. Each time it boots, it starts back at the base. 3) You don't need the actual robot. There is an emulator! https://github.com/pollen-robotics/reachy_mini/blob/develop/docs/platforms/s...
Anyone interested in building/helping build it?
Bob Roswell broswell@syssrc.com 410-771-5544 ext 4336
-----Original Message----- From: vcf-midatlantic <vcf-midatlantic-bounces@lists.vcfed.org> On Behalf Of Neil Cherry via vcf-midatlantic Sent: Monday, December 22, 2025 11:19 AM To: vcf-midatlantic <vcf-midatlantic@lists.vcfed.org> Cc: Neil Cherry <ncherry@linuxha.com> Subject: Re: [vcf-midatlantic] OT: Anyone working/playing with AI professionally? (Update)
On 12/12/25 14:09, Neil Cherry via vcf-midatlantic wrote:
I've taken all I've learned (my brain aches) and started applying it to using AI, in Code under Linux, with a paid Github account. The level of improvement is quite noticeable. This was Jeff's suggestion when he was demoing the MOBIDIC. I also stopped trying to give it a full requirements document and expecting a done product. :-) I've broken it down to more manageable chunks. I've been experimenting with Git repos I control. So screw ups are asy to fix (as long as I use the repos correctly). I've learned that it doesn't pay to give the AI 2 small things at the same time to fix or upgrade. Just one thing at a time. This has gone much better, getting the AI to undo it's changes have been 'fun'. I went round and round in a circular conversation until I could find the correct phrasing it understands. So yes, a Jr. Programmer still. But my coding speed as gone way up.
Now I need to learn about copyright issues with AI code. Seems all the open source communities are very nervous about that. For my simple home projects this is not an issue. Also not an issue at work (so far). I still have much to learn and need to learn about the different AI tools other than the coding LLMs.
Thanks
-- Dave McGuire, AK4HZ New Kensington, PA