EEVblog #586 - Open Source Hardware uARM 4-Axis Desktop Robotic Arm Kickstarter

[Jimmy]

G Dave long time lurker here. I love the vid thank you for sharing.  Keep up the good work and ignore the haters.

[HackedFridgeMagnet]

Well the arm would be good to start learning robotics with. And I dont think it could accidently kill you unlike some of the bigger ones.
6 axis would be better though.

[FrankT]

I've been building mine for a while.  If I ever get it finished, the first task will be to sort my box of through hole resistors.

[electrolux]

That u arm is rather disappointing' I'm not impressed.

[GiskardReventlov]

That u arm is rather disappointing' I'm not impressed.

Agreed, but sounds like the team is young. But it looks like they didn't study what's out there. I have a desk lamp and the mechanism is more stable than this uARM. It's got the arms parallel in the vertical.  The next iteration better look a lot different. They need to study what engineers before them arrived at and understand why. I hate to say it but it's the arduino syndrome. Too much nuts-and-bolts stuff is already done for them and people end up with unclear understanding of what they've built and why it behaves as it does.

[Jimmy]

It looks as if it is not open source yet and now that the kickstarter has been funded and they are selling kits on their website are they going to open source it?

[Jimmy]

Still not opensource maybe someone could take it apart and scan a copy of each part for us?

[FrankBuss]

Still not opensource maybe someone could take it apart and scan a copy of each part for us?

Sorry for reviving this old thread, but here you go, see question 2. This is for the new uArm Metal version, which seems to be a bit more stable than the one Dave tested. But the motions look still a bit jerky too me, and not good position repeatability. And the metal version cost now $339.

A professional robot, like this one, has a repeatability of 0.005 mm:



Problem is it costs $15,000.

I'm searching from time to time for a good low-cost robot, something between the quality of an Arduino-servo-learning robot and the super high precision of an industry robot, and couldn't find anything. But now there is an arm for $1,159, with a repeatability of 0.2 mm, Dobot Magician. Independent review, see 7:12 how it can be used as a pretty good plotter:



And they even claim 0.02 mm for their latest product:

http://dobot.cc/dobot-m1/product-overview.html

But this costs $2,999. Still cheap for the performance, if it is as good as specified. Note, I'm not affiliated with the dobot company, and I don't have one (yet), but it looks cool.

Are there any other good robot arms in the $1k range or below out there? Maybe with 0.5 mm repeatability for $500?

[Barny]

I don't knew why, but 0.005mm and 0.02mm repeatability sound a little bit optimistic.
Even the smallest change of temperature have an effect at this accuracy levels.
Don't talking of the wear of gears and bearings.

[FrankBuss]

I don't knew why, but 0.005mm and 0.02mm repeatability sound a little bit optimistic.
Even the smallest change of temperature have an effect at this accuracy levels.
Don't talking of the wear of gears and bearings.

Maybe that's if the temperature doesn't change too much and not for millions of repetitions, but looks like at least the industrial robot can do it already:

[Barny]

It is able to do that.

The Question is, will it be this accurate under load?

I can't be helped to see this robot as nice little toy

[Dubbie]

Also very very suspicious of the accuracy claims. A big solid cast iron machine tool can only just hit those tolerances.  I think they must be measuring on the exact same path with zero load variation.


Sent from my phone using Tapatalk

[Koen]

As usual : repeatability isn't accuracy, should tend to 0 anyhow, gives no insight on the build quality, is only useful alone to the marketing department and so on.

[FrankBuss]

As usual : repeatability isn't accuracy, should tend to 0 anyhow, gives no insight on the build quality, is only useful alone to the marketing department and so on.

You mean for the Dobot? It is at least as accurate that you can draw pretty clean text with a pen with it. I doubt you can do this with the uARM, the motion is too jerky. And I'm sure the Meca500 is standard industry quality, build quality as good as the big ones (Kuka etc.), only less possible payload, and costs too much.

[Koen]

No, in general. If a manufacturer makes a claim for repeatability but not accuracy, it's crap. Repeatability tends to zero unless you're terrible at what you do and is an easy claim to make for the marketing department but it provides absolutely no value to the buyer as-is. 0.00001mm repeatability and no accuracy figures means you could be wrong by 5mm every time, but always wrong in the same way. And interesting repeatability claims should come with environment and time data. Nobody cares about the repeatability of a new machine in perfect conditions, it should be 0. What's interesting is how it operates after a few months/years or in its potential production environment.

[HackedFridgeMagnet]

Nobody cares about the repeatability of a new machine in perfect conditions, it should be 0.

What machine is that? I'll buy it. 

[H.O]

No, in general. If a manufacturer makes a claim for repeatability but not accuracy, it's crap.

In my limited experience it depends... Having a bit of experience with KUKA robots I have some literature right in front of me (general sales brochures and technical documents for the specific robot I'm currently working on) and neither of them state accuracy, only repeatability. I wouldn't exactly call KUKA crap....

I'm sure they DO have accuracy figures available but it's not what they're specifying in their sales brochure or in the 26 page technical data sheet I have here.

Now, if you're talking about machine tools (mills, lathes etc) then it's another thing all together. There absolute accuracy means a lot more than repeatability.

[Koen]

It's in extra documents because it depends on the region in use, regions crossed, movement sequence, application, load type, load weight and more.

This is just my usual, simple and boring warning for people shopping for their first CNC, whatever it is : repeatable doesn't mean accurate.