PDA

View Full Version : Retractable Arms



Blinky
02-07-2013, 02:13 PM
Hi all,
I'm looking for suggestions here. I'd like arms to be able to retract into the torso of the robot. They don't have to retract all the way in. They also need to be beefy. Roughly human dims. Being able to hold say, 3.5lbs. The wrist and claw assemblies alone weigh along those lines. Other than that, fully articulated as the arms available here at Trossen. Any ideas? Power is not an issue. This is a large robot. Would like to use servos. Thanks!

Gertlex
02-07-2013, 04:13 PM
I would suggest retraction by having them fold down into cavities in the main body of your robot... nice and simple... Doing like a turtle, on the other hand, would be a nightmare probably.

KevinO
02-07-2013, 05:05 PM
Take a look at the WidowX arm design Trossen provides. It folds up nicely. Though I should warn you would probably need an arm with MX-106s or such 3.5lbs fully extended is a lot of weight. Take a look at the weight charts that Trossen provides on their arms. It'll give you a good idea where you need to go.

jwatte
02-07-2013, 05:54 PM
An alternative is to use geared motors with encoders and your own controller. Basically, building your own servo. The benefit is that you can build your own mechanical armature to move the arms, and make your own speed-vs-power trade-off, with whatever leverage you need. Need a lot of holding torque, but not much speed? Use a lead screw!

tician
02-07-2013, 07:27 PM
I would think an MX-106 would be around the absolute minimum even if you added a counterbalance system for the weight of the arms. I second jwatte on the lead screw / worm drive / linear actuator if you can afford the slow speed and complete lack of compliance/elasticity without modification (springs/elastic bands linking actuator to arm would give some safety for both bot and bodies). Another option would be the dynamixel pro, if you can wait for whenever they get released. Last I heard, they use cycloid gearheads, so they would share worm drive's inability to be backdriven (compliant).

Assuming my brain stops screwing with me, I'm hopefully making an order for some banebots P60 gearheads, RS-390 motors, capacitive rotary quadrature encoders, and some motor driver boards for darsha deux (misumi and newegg orders as well, I hope). If you pair the right motor with a high enough gear ratio, you are supposed to be able to get up to ~30 lbf-ft before destroying the gears (about as much as the biggest dynamixel pro for a fraction of the cost, assuming the time you spend getting them working as servos is not worth all that much). Their P80 gearheads can go even higher - food for thought.

Blinky
02-08-2013, 10:09 AM
Thank you kindly for the suggestions. Unfortunately, the design of this particular robot dictates a "turtle" like retraction rather than fold of the arm(s). That said, they don't actually need to retract "inside" the torso necessarily, but rather into a compact position that would give the illusion. Not that this makes it any easier.

rebel
02-08-2013, 11:30 AM
Use the design from Wall-E, would not be overly complicated to make, or get to fold into its self

KevinO
02-08-2013, 11:46 AM
I'm sure you can figure out something on the design but I'm curious about how you will handle the weight issue? The cheapest arm setup that can I've seen that can do 1.5 kilo or higher is $6,000.00 US. That's a big investment.

Blinky
02-08-2013, 04:01 PM
https://vimeo.com/59266990
Here's a crappy vid of what I'm doing now. The claws are attached to the end of a rubber rod and go in and out on a liner slide. There isn't a lot of clearance once the torso is in place. I'll figure something out but not for $6,000!! Yikes. Maybe some kind of telescoping action? Aluminum framing. The most robust servo available. These arms don't actually have to "lift" anything other than themselves. This is strictly animatronics.

jwatte
02-08-2013, 05:38 PM
These arms don't actually have to "lift" anything other than themselves

If you can suspend the arms using linear bearings, and just move them in/out with some other mechanism, then a simple DC motor with limit switches may be the perfect solution. Let the linear bearings take the brunt of the force.