I’ve been looking at the latest Robosapien V2 design and there are a lot of things to like about it. The price point and relative simplicity of design compared to all the things it can do is obviously the coolest thing about it…
However, the most interesting technological accomplishment to me are the articulated hands that the Robosapien has.
They not only look terrific they work by way of motors in the robot’s chest that essentially push or pull the fingers closed and open and have sensors to determine whether something is in the hand or not. Very cool… I’ve been thinking of making some basic expansions to the design to make the index finger operate independently of the others (esp. for precision pointing), upgrading the strength of the grip, improving the responsiveness of the sensors so that the hands can dynamically determine how hard to grip and maybe having the thumb rotate outward so that the palm can be flat…
I wish they sold those hands as a standalone item ;o)
i once made a simple hand out of foam board and nitinol (that memory shape wire). it might be something to look into as it takes far less space than any motor assembly might.
That’s a pretty good idea. I’ve never messed around with memory wire or synthesic muscle but it definitely seems like the future…
Nitinol or the other types I belive there are 3 total, are cool to mess around with but not that stronger to make a practical application. Now there are air muscles that contract and expand with air pressure. These are pretty good but you need an air supply that can give ~ 50psi if I remeber correctly. And they can provide a decent amount of lifting power.
those air muscles are very interesting but getting a small air supply inside a small bot just seems too difficult. i used multiple lines and a series of pulleys to deal with the strength issues inherent with the wires . i had three bend functions: index, thumb (fixed opposed), and the 3 remaining fingers. in addition 2 servos power the wrist through a pull-pull cable system. The whole setup was designed to fit within the forearm but I never got past initial testing and the foam board was far too weak to support any kind of weight (the strength and heat of the wires eventually cut the arm to bits, I guess I should have realized that I was making a hot wire foam cutter )
I was playing around with eMachineShop and came up with this rough design:
I think I might revise it and remove the finger groves and go more the robo saphian route, where the fingers ar mounted to the flush palm area.
Here is the finger made of “links” that I designed. You could use the muscle wire to move them. I could not render these links because eMachineShop will only allow redering of one part at a time. I can render one link when I have more time.
Wow, very cool Mike. The finger looks great. While it would be cool to make all of the fingers functionally distinct I actually think that might be overkill. I mean really, how many times to you really need your middle, ring and pinky fingers to be functionally distinct. Maybe making that Hook 'em horns sign Not having to have separate controllers for all of those would save a lot of space as well.
As far as the hand portion, I was thinking of somehow geting that thumb joint to be alternately opposible or in line with the palm (like human hands are). There are a lot of situations where you want you hand to be flat against a surface… Anyway, nice work…
I’ve seen an application of air muscles to make a full-sized hand (I believe it was at the MIT website), but I’ve never seen small enough air muscles, nor small enough pneumatic controlelrs, to suit your needs.
Then again, I haven’t actually looked for them.
Clippard Minematics might be a good place to look, since they specialize in small pneumatic parts (for the controllers, that is; I don’t believe that they sell air muscles).
As to the problems with a small air supply, couldn’t one just use a 12g CO2 cartridge with a regulator?
I havnt really worked with wires but wouldnt you need two sets for each joint. One to pull the other to push? You turn it on the hand closes how does it open up?
What I was thinking of doing is run a spring through the holes of each link and then thread the muscle wire through the spring.
If you take a strait spring that pulls (normally compressed) and bend it, when you let go, the spring snaps back strait. It would not be strong, but should be strong enough to move the fingers back to a open posture.
Here is a picture of a machineable hand with fingers. I think I might revise the design so the thumb will be more in the middle but I am not sure. Any opinions?
wow I didnt know emachineshop could do that. they must have updated the software since the last time I used.
Anyway can I make a suggestion? Purhaps angle the them towards teh other fingers by 20 degrees or so. Not to say anything is wrong with your design I think it would just make it look a little more realistic.
I agree with your opinion. I tried to angle the thumb about 20 to 30 deg. but the problem I encountered with eMachineShop was adding the drill hole to match the angle. eMachineShop does not allow angled drill holes, or at least I have yet to figure out how to do it.
I might add the thumb in the middle underneath the palm more like the robosapien hand example. This way, the thumb would have a better chance at gripping an object. I have designed literally several dozen models trying to get it to pass the strict design rules that eMachineShop uses.
I know I have designed several bot parts in the past, and have not made anything yet, but I figure since I can’t do anything right now, I have the time to carefully design each idea I have and then archive it for latter use.
So far in my archive I have:
-A backpack cover
-Hand with finger links (Still a W.I.P.)
I have even tried to design my own humanoid chassis but have put that on hold until I can learn from a Lynxmotion biped design. I do plan on using my sensor skull and later down the road the hand design.
If I may share my opinion again. Dont use e-machine shop. The free software thing is nice but it is very limited. Not only that but emachineshop has some very steep prices for onetime parts. When ever I have wanted something machined I draw it up in Solid Works and go to www.cnczone.com and post it in the RFQ forum. Theres a lot of good people over there who can do the job for your for a cheaper price and sometimes better quality as well.
Also you dont neccessarily need a CAD program either. Depending on your drafting skills you can draw out your parts in a 3 view, isometric template and take it to a local machine shop and they will probably have better prices too.
That sounds like something I may need to research. I like the ease of use of eMachineShop’s software. I hope other cad type programs are just as easy to use.
Compared to Emachineshop, Solid Works is incredibly easy. I havent used EMS in a while so im not sure exactly what thew newer versions are like but SW is the easiest program I have ever learned. This coming from someone who knows Catia V5, AutoCAD, EMS, TurboCAD, and a little bit of Pro E.
Well If I am able to master eMachineShop [size=75](I’ll call it ems for short)[/size] on my own, then SW should be like cake work.
I wonder if I can import ems files to SW. There’s a DFX R12/LT12 export feature with ems. would that work with SW?
By the way Topher, thanks for your input.