Remove this ad

Lead

Aug 5 16 6:50 PM

Tags : : ,

Hello all, 

My name is John Cheshire and I'm a 3D designer certified in Solidworks, as the Ada hand was developed using blender this poses a few problems importing into solidworks, whilst you can work from STL files it's really not worth the hassle, so i'm going to start from scratch and try to eliminate as many of the hardware challenges you've listed along the way, such as silicone grips, adjustable tensioners etc, I will of course try to maintain the sleek look of your original design. I own two ultimaker 2s, a form 2 and a form 1+, so I will be posting updates of my progress as often as I can. As of yet I haven't purchased any of the components and for the time being I will be working off the provided dimensions in PDF's, any infomation and other dimensions that could be provided would be greatly appreciated, as well as any one elses ideas or input are most welcome.

Looking forward to working with you all.
 
Quote    Reply   
Remove this ad
Remove this ad

#1 [url]

Aug 8 16 7:46 AM

Hello all, this weekend i've put a couple of ideas together, firstly implementing a middle section that allows you to slot on ABS/PLA finger sections and also silicone grips, doing this will also greatly reduce the amount of Ninjaflex you have to print with, this will allow you to print the entire palm in abs/pla also, which should bring costs down. The ninjaflex skeleton was designed so you can print it flat down without need for supports. I also added a tensioner idea on the end of the finger tips too, nothing has been tested as of yet, awaiting the arrive of some ninjaflex for my printers. Welcome any feedback, ideas and concerns.

In the meantime you can download the pdf design here (just a quick concept model)
https://view.publitas.com/p222-10736/left-hand-index-pdf-1/
 

Last Edited By: JohnCheshire Aug 8 16 7:55 AM. Edited 1 time.

Quote    Reply   

#2 [url]

Aug 8 16 10:09 AM

Hi John,

This is a fantastic start, nice one! Definitely agree with your approach, working from STLs isn't worth it. 

Firstly, the "floppiness" (technical term) of the fingers will depend on the width, height, and thickness of the Ninjaflex in the joints. If you want a starting point for "calibration" I'd recommend that each of the two struts that make up each of the proximal joints are 1.2 mm long (along the finger axis), 6 mm wide (along the bend axis of the joint), and 2 mm thick. In the same order try 0.7 mm, 6mm, and 2 mm. The actual floppiness will depend on your printer and print settings.

I like the idea for the tensioner. However, I worry that the Ninjaflex side of the ratchet will flex and allow it to slip. The tendon is under a surprisingly high tension! Definetly worth a print to make sure though.

Finally the angles of your joints and the diameter of the tendon channel all look good. It's really encouraging to see someone take on all the design challenges at once with an integrated approach. 

Good luck and I look forward to seeing your developments. Please feel free to message with any requests for dimensions.
 

Last Edited By: JonathanRaines Aug 8 16 10:29 AM. Edited 1 time.

Quote    Reply   

#3 [url]

Aug 8 16 11:33 AM

Hi Jonathan,

Thanks for the response and feedback! 

I myself have never used ninjaflex, I mostly work in ABS, the closest thing i've probably used to it is toughened resin on the formlabs printers, so alot of the dimensions were just rough guesses looking from the original design. Thank you for the dimensions though, thats a massive help. Seeing as the demensions of your struts are 6mm wide and there are 2 of them that puts you at a width of 12mm not even including the gap between them, which is already wider than my skeleton base (10mm wide), I could either go down the path of making it wider by 2mm with no gap between the struts, or alternatively keep my width of 10mm, get rid of the gap completely and increase the thickness incrementally untill I hit the sweet spot, looking at the space i've got increasing the thickness is probably my best bet.

I agree, even when I was designing the tensioner I was skeptical, still it's an idea to build from, have you guys had any luck trying ideas for different tension methods? The size of the parts seems to be the biggest challenge, FDM priting certainly does have its limitations.

Again thanks for the response, i'll be posting updates of my progress through out the week.

Quote    Reply   

#4 [url]

Aug 10 16 6:53 PM

After measuring the Current ADA finger joints from the .stl file using solidworks, I discovered that these are the closest measurements: 0.8mm long (along the finger axis), 5mm wide (along the bend axis of the joint), and 2mm thick, at least this was the case with the LH Index finger joints. So i'm going to use that as a starting point, below is a pdf with a comparison of your current joints vs my own, the ninjaflex arrived today and I intend to print the index finger from ada v1.1 and also my own skeleton design and do a "floppiness" comparison. Without a motor or board I don't know what the best way to test this would be, perhaps with weights? Or alternatively I could go ahead and order the parts. Any advice on print settings for the ninjaflex? I'm using a Ultimaker 2 with simplify 3D software, any cura settings would also be fine.

https://view.publitas.com/p222-10736/drw1-pdf/
Any thoughts and concerns are welcome!

Last Edited By: JohnCheshire Aug 10 16 6:56 PM. Edited 1 time.

Quote    Reply   

#5 [url]

Aug 15 16 1:37 PM

I've done some more modelling on the skeleton section of the hand. See PDF linked below.
I realised that it's probably best that I modify one of my ultimakers before going ahead with the ninjaflex testing, because of this I decided to make my model equation driven, in the pdf the is a list of values tied to variables and equations, if I were to change one the values the model would be able to calculate and rebuild the model without it breaking, so this means I don't have to start from scratch if I need to change a dimension at any point, it will also allow me to proceed designing without testing all the time.
If anyone has any thoughts on the dimensions, for instance: the angles between fingers, I would love to hear them. I'm also alittle perplexed as what to do with the thumb at the moment.
I noticed that the motors on the current ada hand don't run exactly colinear with the fingers and are angled a few degrees off, i've managed to align the motors in my model with the fingers, this will hopefully improve the linear performance of the fingers, and take away unnecessary tension. The motor for the thumb i'm not 100% sure where to place at the moment, don't know whether i'll be able to get it colinear with the thumb, is there some kind of equation you can use to work out the thumb angles? Or is it a matter of trial and error? - I can probably run a simulation for it in solidworks.
Also regarding the motors, in your model are the motors fully extended? If not, what length are your motors able to extend to?

Just a few notes:
The middle finger is running verticle. (perpendicular to the wrist)
The shape of the hand isn't final, just a random spline I made for perspective
https://view.publitas.com/p222-10736/master-drawing-pdf/

Last Edited By: JohnCheshire Aug 15 16 1:45 PM. Edited 2 times.

Quote    Reply   

#6 [url]

Aug 17 16 9:08 AM

Hi John, 

An equation driven hand is a good idea. Just be sparing with fillets and alike as I've found they often break if the variables move too much from the values that you start with. The angles between the fingers look good, we use 5° too. However, consider rotation around the axis of each finger as well. If you look at your own hand, you'll notice the bend axis for the fingers aren't co-planar. Instead the fingers close inwards. This is important for enclosing objects properly. 
The thumb is a hard one! The motion of a human thumb is very hard to replicate. The important part is getting it to move across so that it can oppose the fingers in both a whole hand, tripod, and pinch grips. I generally do this by simulating the bend. 
With regards, to the motors, they are fully extended in the model. You may also find this datasheet useful: https://s3-us-west-1.amazonaws.com/firgellidownloads/PQ12+Datasheet.pdf

Kind regards, 

Jonathan

Quote    Reply   

#7 [url]

Aug 17 16 11:37 AM

Hi Jonathan,

Yeah I noticed that with fillets too, the only fillets I have in the current model are on the skeleton finger tips, I made those equation driven too, they're governed by the skeleton finger width ( M/2 ), but i'll definitely watch out for that. I'm alittle puzzled about what you are saying for the bend axis of the fingers, currently the bend axis for each joint is perpendicular to the axis of the finger (see below), are you sugesting that perhaps some of the the joints should be parralel to the wrist or some other angle to make them bend straighter? If you have a visual example I may understand better. I guess i'll just have to build a simulation and test the thumb then. On a side note my friends and family are concerned about how much i've just been staring at my own hands recently haha.
Thank you for the datasheet, luckily they're infact the dimensions i used.

https://view.publitas.com/p222-10736/finger-bend-axis-pdf/

Last Edited By: JohnCheshire Aug 17 16 12:54 PM. Edited 2 times.

Quote    Reply   

#8 [url]

Aug 18 16 10:46 AM

Okay, I think I understand what you're saying about the fingers closing inwards, here is currently how my fingers would simulate when fully enclosed:
https://view.publitas.com/p222-10736/simulationtest1-pdf/
So in this simulation there's alot of space between the fingers, whereas when you make a fist with a real hand the fingers close towards each other leaving no or minimum space, and this can be achieved by changing the angles of the bend axis so they're not perpendicular to the fingers, I will go back and try to resolve this.
If i'm wrong about anything above let me know.
Thanks for pointing this out.
 

Quote    Reply   
Remove this ad

#11 [url]

Aug 18 16 4:42 PM

Right I'm with ya. I will set up 2 configurations and post my results, have a feeling my way may will put the motors under more strain, but I can't know that for sure yet, maybe a combination of both our ideas would work. Haha, not quite at that stage yet, though I have spent most of this week measuring my own fingers with a vernier. Where are you guys based by the way? I think I read Bristol somewhere.

Quote    Reply   

#12 [url]

Aug 22 16 1:20 PM

Ran some simulations on the weekend, after a lot of messing around with the angles and finger rotation these are the dimensions i've landed on.
https://view.publitas.com/p222-10736/master-file-v2-open-close-sim-pdf/
The PDF also shows what the fingers look like fully open, half closed and fully closed. At first i just tried rotating the fingers but then I spent some more time staring at my own hand and came to the conclusion that the fingers are rotated due to curvature of the back of the hand, so i tried to replicate this by cutting the skeleton into 4 pieces with angular edges, when they come to sit in the casing for the back of the hand they will be angled, hence rotating the fingers, unfortunately this wasn't enough and the fingers ended up spreading out unaturally when closed so I then ended up changing the angles of the axis that the fingers bend alond. Anyway this is where I am now, unless anything looks wrong i'm ready to make a start on the thumb.

On a side note i'm thinking about ditching the tendon channels on the skeleton part and instead runing them inside the ABS/PLA fingers, pretty sure there will be less friction this way and the holes would probably print up better due to the orientation the fingers.

Last Edited By: JohnCheshire Aug 22 16 3:09 PM. Edited 1 time.

Quote    Reply   

#13 [url]

Aug 25 16 10:18 AM

I'm going to start printing and testing some of the parts i've modeled before I push any further. So I may not post anything for a few days.
Once i've got some solid results I'll be back. Considering purchasing a Lulzbot mini with a flexistruder tool head, instead of upgrading my ultimaker, have you guys got any experience with that printer? It seems to have decent reviews.
 

Quote    Reply   

#14 [url]

Aug 25 16 1:02 PM

We've got a Mini but don't have the flexy struder mod. It was a good workhorse for about a year but is showing its age now (we do use our printers heavily). If you go down the Ultimaker modification route, I'd recommend the Flex3Drive. Be warned, Ultimakers print beautifully out the box and it can be hard to get them back to pristine condition after mods. We have one dedicated to flex and two untouched for PLA. 

Quote    Reply   

#15 [url]

Sep 10 16 2:03 PM

Hi all, still hard at work and still doing test print for my design so far, struggling alittle with the ninjaflex at the moment, but I think i'm getting the hang of it. Designs in my full time job are taking up alot of my time so progress has been slow. Will try and get an update for you all soon. Thanks for the mention in the newletter by the way!

Kind Regards
John Cheshire

Quote    Reply   

#16 [url]

Oct 2 16 6:28 PM

Hi all, I've been running some basic tests on my current pinky finger design. The finger works as intended, bends easily and returns to its form, however I think the flexible skeleton needs to be a tad thicker for a better return and also to hold its shape little better when relaxed. Also, currently there are 3 steps on the back of the flexible skeleton to stop each part of the finger from sliding off the skeleton, in place of this I might make a pin system, trying to push the skeleton through the fingers was a bit of a nightmare, a pin system should make assembly easier.
More updates soon.

Quote    Reply   
Remove this ad
Add Reply

Quick Reply

bbcode help