In case I’ve put together an AR4 from Annin Robotics, and it was great fun, a good learning experience and then changed the code to be able to control it in various ways. the main issues with most arms is the payload and reach which is quite low for most… The AR4 has a reach of just under 700mm and just under 2kg of payload and you can easily get to its limits.
Now it’s also been sitting on my desk doing nothing for the past 3 months (got it for about 9-10 months now)
But I played around adding pneumatics etc to it and it worked fine. You sould check that website out!
To throw in my $0.02, scoping is usually easy enough once one is clear on the deliverables. e.g. if you want something to flip burgers, is it just quarter pounders or do you also want it to be able to do half pounders also? Payload, dof etc then comes out of that. If you want a motion control rig for movie SFX, then accuracy, repeatability, range, payload etc. will come out of that and the kinds of shots that you want to be able to do. When you have deliverables figured, then assess other options. e.g. adding an A axis to a CNC might be a better option if precision milling and sculpting is a requirement.
It sounds that you don’t have a particular use in mind, and the main thing you want to get is to learn about robotics etc. I’d think the suggestion of an open source system would be the way to go then. Also check out modular parallel robots like the ones here - https://youtu.be/3fbmguBgVPA. The components can be configured in different ways as you’ll see from other videos on that channel.
