Robot arm musings
Florian Festi wrote 05/09/2021 at 08:52 • 2 pointsI don't have a robot arm. I don't need a robot arm and I don't have the time to build one. Damn, I have enough other projects and problems I should not even think about robot arms.
Nevertheless I always read through any robot arm projects that come up - here on hackaday or else where. There are quite a few very impressive robot arms out there. But most of them are either machined or 3D printed. I am more of a laser cutter guy. But I am just too aware that an robot arm made from materials suited for a C02 laser would probably be just not stiff enough. Yes, many robot arms have position sensors directly on the joints to compensate for flex and backlash in the drive train. But if the arm itself bends that's not helping much. But what if we could measure the position of the end-effector directly and then compensate for any errors in software? What if we could use the same method as four-wire-sensing for resistors: Separate the load bearing from the position sensing parts? Just have a second arm inside or beside the fist one that is only connected to the base and the end-effector and has angular sensors in the joints.
To be honest I was pretty proud of myself and the idea. But whenever you are too proud of your idea there two questions to ask:
Why isn't everybody doing this already? and Hasn't somebody tried this before?
For the first one I guess that having a wobbly robot arm is just not great for many applications even if you can position it precisely - after the wobble has died down. But may be that's a price some are willing to pay - depending on the application.
For the second one: Well, there is a this paper: "Independent load carrying and measurement manipulator robot arm for improved payload to mass ratio" that can be found at https://www.researchgate.net/publication/324654377_Independent_load_carrying_and_measurement_manipulator_robot_arm_for_improved_payload_to_mass_ratio
*sigh* They even make the four-wire-sensing analogy...
But they only have tested a toy robot arm with two joints. Results look good, although I guess there are several important properties they just did not test. So there is still work to do and fame to be earned.
But as I said "I don't need a robot arm." But I would love to read about it.
share
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
There's a "Redefine Robots" challenge in the HackADay Prize now. Just saying...
Are you sure? yes | no
It certainly has been done in giant brick laying robot arms. They can't be made ultra-stiff since they need to be light enough for easy transporation, and are exposed to wind in addition to the more predictable sag from payload weight and vibration from the arm's own movement. But I haven't seen any desktop size arms use that approach. Hard to say whether it would be a good solution for them, since I've never seen a desktop arm do any useful work in the first place. Brick laying is a very specific application, so you can really optimize the compensation system for it.
Are you sure? yes | no
Yeah, that makes a lot of sense. Building robots can't afford to overbuild by a huge factor just to get the necessary stiffness.
Yeah, desktop robot arms more or less by definition are not doing anything useful. Otherwise they'd be working in some factory or lab and not be sitting on a desk top. The professional robot arm are still too expensive to casually sitting around on a desk and the cheap ones can't do much. I'd hope that this technique may get the cheap robot arms a bit of a boost. Half a Millimeter accuracy even under several kg load seems quite impressive - even if only achieved statically or with very slow movement.
Are you sure? yes | no