I'm designing a 3D printed robot arm that doesn't suck.
It uses all brushless outrunner motors for at least the 4 large axes (final two in wrist may be hobby servos for now).
The arm is similar dimensions to a human arm. Initial calculations show it will be able to lift more than 2kg at full extension if I can make it strong enough. Actually the (rough) calculations say more than 10kg but that would break something. It's also 200 milliseconds for 90 degrees of shoulder movement, supposedly.
All hardware and software is open source, including the brushless drivers.
It uses low cost 3D printers. I am currently using a $320 Monoprice Maker Select modded with a 1.2mm nozzle and I will be adding a $450 TEVO Black Widow modded with a 2.5mm dia nozzle to the mix. I may try a pellet feeder. The goal with the large nozzles is to increase strength and reduce print time so it's not maddening to print. This will make it easier to iterate.
My goal is to make an open source low cost arm that is useful for manufacturing, and then design open source workcells for it and actually use it for productive work.
Cool. The onshape link asks me to login right away. Without logging in, I don't see your 3D item. You're invited to also upload the robot arm to www.3dprintmakers.com.
Onshape is free and cloud based. It's good because you can easily create an account and edit the file yourself, but a bummer because it's not free software and you can't export raw files, only dumb solids.
Despite the drawbacks, I'm keeping it in OnShape though, where anyone can fork and edit the file for free.
will it be fully 3d printed (no metal/carbon tubes)? I think the main problem with hobby robot arms is flex in the arm itself and backlash in the motors.
Low cost arms (eg. dobot) have "0.1mm repeatability" with no load but the moment you put a kg on the end effector it starts flexing.
Aside from bearings, motors, and possibly some fasteners, it will basically be entirely 3D printed.
I'm using printed spur geartrains that will have all kinds of backlash. In an effort to make something that is primarily printed, I'm throwing any hope of precision out the window. At least mechanical precision - we can use visual servoing to get more accuracy at the end effector.
But my hope is that the other benefits (low cost, easily changeable design, good speed and strength characteristics) will make it useful even with poor rigidity.
I imagine for example pulling things out of fixtures and moving them in to boxes.
There must be a good variety of tasks that can be done with an arm like this. I'm working to the strengths of 3D printing and not trying to fight what it isn't good for.
That said, I may reinforce the frame with a carbon fiber wrap if more stiffness is needed.
The goal is to have a low BOM cost AND low tooling cost so I want to avoid the need for any metalworking equipment.
It uses all brushless outrunner motors for at least the 4 large axes (final two in wrist may be hobby servos for now).
The arm is similar dimensions to a human arm. Initial calculations show it will be able to lift more than 2kg at full extension if I can make it strong enough. Actually the (rough) calculations say more than 10kg but that would break something. It's also 200 milliseconds for 90 degrees of shoulder movement, supposedly.
All hardware and software is open source, including the brushless drivers.
It uses low cost 3D printers. I am currently using a $320 Monoprice Maker Select modded with a 1.2mm nozzle and I will be adding a $450 TEVO Black Widow modded with a 2.5mm dia nozzle to the mix. I may try a pellet feeder. The goal with the large nozzles is to increase strength and reduce print time so it's not maddening to print. This will make it easier to iterate.
My goal is to make an open source low cost arm that is useful for manufacturing, and then design open source workcells for it and actually use it for productive work.
You can look at the jumble of CAD BS right now. https://cad.onshape.com/documents/5b474270e4af0ef979e6fade/w...
Find the tab with the assembly called "Arm3 Assembly".
Please fork it and contribute.