PDA

View Full Version : [Question(s)] Biped ankles, & The hand of Zychor - WIP



Slugman
07-12-2011, 11:11 PM
Hi everyone. After converting my 22 DoF AX-12 hex to a biped, I have come up against the old "Ankle servos getting too hot" problem. I have worked out a simple hinge & lever design that would reduce my available foot tilt to + or - 20 degrees & increase the holding torque significantly, but that leaves the lever only connected to the servo-horn by one 2mm bolt. Basically, the servo pushes a hinged lever down (23mm centre to centre), which is connected to the foot, while the foot tilts on a hinge set 16mm off to one side, & approx 23mm below the centre of the servo-horn.

Apart from the increased 'slop' due to the tolerances in the lever & hinge connections, is there any point to me bothering to make it? I.E. Have any of you done this already & found the problems?

I have access to a lathe, so if the connection to the servo horn needs strengthening, then I should be able to make up a little plate to allow all four servo-horn bolts to be used, with the lever bolting seperately to the plate, but I guess my question is more about the basic mechanics of the system - Will it, or 'should it' work? - I am trying to play "Devils advocate" to make sure I have covered all potential issues. :confused:

The reduced range of movement should be more than sufficient for walking on flat ground, while the available 20kg/cm holding torque when the foot is level, & much higher when the foot is at 20 degrees tilt, should prevent the servo overheating. Disclaimer - My torque calc's may be wrong! The foot doesn't need to be tilted fast, so lack of speed shouldn't be an issue either, & it doesn't make it that much taller compared to the normal plastic brackets. The extra weight will be low down, & will only be a few grams.

PS DarkBack - Yes, I know that it really means that my robot is too fat......... :tongue: I WANT to carry more uneccessary weight, while being able to maintain static stability in the entire stepping motion. I hate shuffling, so replacing the servo with a hinge & two opposing springs is not my preferred option.

darkback2
07-13-2011, 12:33 AM
PS DarkBack - Yes, I know that it really means that my robot is too fat......... :tongue: I WANT to carry more uneccessary weight, while being able to maintain static stability in the entire stepping motion. I hate shuffling, so replacing the servo with a hinge & two opposing springs is not my preferred option.

Who said your robot is fat?

Currently I have tried to take the load off of my bipeds ankles using springs...Levers are always an option (especially if they work!)

Seams like the only way you will know for sure is if you try. Who knows maybe we will all be using the "slugman levered ankle design" by this time next week.

Good luck.

DB

Slugman
07-13-2011, 05:34 PM
Yeah, I looked at using a spring on the outside of the foot to take some of the weight, but thought the lever might be a better option with a similar weight penalty. I was just hoping someone had already done it & could give me a few pointers on what they have already learned. Dunno about next week - Given how long it took me to make my prototype mech warfare machine gun, it could take a while.... Not so sure about trying to claim the design either. I am pretty sure levers have been used once or twice before in the history of robotics. :veryhappy:

FYI, the gun prototype came in at over 400 grams on its own, & I have been too worried that I might blow it or the Arbotix up, so I haven't tried to get it to work - Need to test controlling cheap micro-servos first, before I try with the high-amperage motor & speed controller. I ended up ordering three of the good gearboxes on 16th June instead - Hope they turn up soon.... Dang snail-mail. :genmad:

Kar
07-14-2011, 03:34 AM
Hi Slugman

I dont know much about bipeds but if you could post some pictures of you Ankle servo set up with changes and with out. I will look over them and see what we can come up with. Im prity good with this sort of thing. Not realy shaw what your trying to do but i hope some pics will clear this up then i can offer some thoughts to what could be done.

Slugman
07-18-2011, 12:30 AM
Well, I'll try to attach a picture of what I was thinking of making - the sketchup drawing I made to work everything out. Currently the weight of the robot is supported by the join between the two servos, so I might have to change where the hinge plate connects to the ankle, so the weight is more evenly distributed.

Anyway, since I have been too slack to start cutting the parts out yet, comments, suggestions & critiques are more than welcome. :happy: At the moment, all I am sure of is that the max & min angle of the foot is enough for walking. I am going to cut the new parts out of 1mm stainless & bend & drill them. Yes, it's a little heavy, but I need the strength.

Kar
07-18-2011, 08:31 PM
Ok thanks for the pictures. Now that I can see what you are try to do.


Firstly the hinged idear. I realy dont think this is going to work. The reson why is beause you are making your servos work like a crank pushing a con rod. From what I under stand about bipeds is, that you are a lot better of keeping to the center's of your servos. Because you are no longer using the center line as a tiping point you will start to have other problems I think down the track. I think one of the bad sernarios could be rather then the leg tilting to the side the foot might lift up on the out side.
It also gives you to many hinged points. The more hinged points you have the more problems your going to have. Specily when there not central hinged points.


Ok I recon the best thing to do and you have probly been told this by other's is use bigger servos.
Sadly I have just change my desktop from window to linux and I dont have any good software to ezely draw up some iders for you to see what els you could try.


Ok I hope that makes sence for you. I do suffer from dyslexia and if your not shaw what that is google it. LOL


Thanks
Garry

Slugman
07-19-2011, 06:36 PM
I wouldn't worry about the spelling too much - It's certainly not as bad as some that post on here, & yes, I understand. :happy: It's one thing that got drummed into me as a kid - Listen to the message, not the messenger.
Good point about the foot tilting rather than the leg, although I was thinking of making the hinge on the outside of the foot, not the inside, so there should usually be more weight over the crank side of the foot. I might have to make the foot wider, so the hinge is closer to the centre line of the foot surface.

I understand your point about moving the hinge away from the center line of the leg, but I think I will try to make this anyway, just to try it out. I should be able to adjust the walking sequence to account for the change in the way the leg tilts, as I am using poses, not Inverse Kinimatics.
I am currently working out how to draw up a file to send away to a CNC / Water Jet company to cut out small bits of steel for another project, so I think I will add these in since the extra cost for the ankle parts would be quite small compared to the total cost to get the other items cut out. :happy:

Slugman
09-05-2011, 01:24 AM
Hi everyone. As you might remember, I am working on a changed ankle & hip design to try to ‘fix’ the overloading of the servos for bipeds. Unfortunately I have run into a problem I didn’t foresee, & was hoping for a little assistance – I need a .dwg file format for the laser-cutter business before they can cut my design out of 0.8mm stainless. I wanted to share the file on Trossen, as I have nested the ankle parts with all the parts required for my experimental hand – It uses the same basic principle as the MechaTE hand, but in AX12 scale with moving thumbs… :) I have been calling it “The Hand of Zychor”, since that’s the name of my ‘bot & I have delusions of grandeur. >:P It should work in any configuration, from two to five fingers.

I drew everything up in Google Sketchup & used a free CAD program called DoubleCAD to open it, then saved it to a .dwg file. Unfortunately the CAD program has cut lines out of the original – According to the Laser cutting company down here it is “Heaps of them” & although I only found two large ones, they could be right, as the CAD program is really abrupt with zooming in – One click on the mouse wheel & you are close, & one more click & you are past where you wanted to be. I am afraid I am not good enough with CAD to work out what the %&^&* I am doing with it! Unfortunately the program was free, & I seem to be getting my moneys worth. *grumble grumble, whinge* It’s not their fault, but definitely mine. *sigh*

Could I please beg, grovel & generally plead with the many CAD experts here to have a quick look at it & let me know just how bad it is? & if they could fix it, I am more than willing to make whatever they want in Sketchup as payment, as I am not too bad with it....

elaughlin
09-05-2011, 07:37 AM
Ill take a look at it. It does look pretty strange to me. It seems to have one continuous line through all of the parts... and like you said, some lines are missing. I use autodesk inventor so I can make it into a new dwg or dxf file if you want for the cutting service if you want to send me the CAD file. Can you convert it into a STEP file or anything like that?
3460
This is what it looked like in the Autodesk DWG Viewer program.

Slugman
09-05-2011, 06:30 PM
Oh, I thought that since it was going to be cut out on sheet metal that they would need it to be in one piece, so I removed the lines of each individual part while it was in sketchup. I wasn't sure if the lines would confuse the laser cutter or not - I.E. Didn't know if the cutter would try to cut along the line, so I removed them. Unfortunately the conversion between sketchup & .dwg removed some lines as well. :sad:
I'm at work now, so I will have a play this evening. The image you posted looks practically identical to what I get to see with DoubleCAD. I'll put the lines back in sketchup at lunchtime so the individual bits are complete, then tonight I will see if I can change it to a STEP file. Thanks for your time. I really hope this works - Both the ankle & the hand!

OK, just had a few quiet minutes at work, so the sketchup file is now fixed - All the seperating lines are back in.:happy:

Slugman
09-06-2011, 05:07 AM
Nope, no good - I have the option of saving as a STEP file, but it says that there are no solid objects. I tried using the "Push/Pull" tool in sketchup to make one of the objects 1mm thick & re-opened it in DoubleCAD, & it has appeared in the program as being correct, but it still won't let me save it as a STEP or STP file due to the lack of solids (I have both options).
Other options are;

2CD, TCW, DWG, DXF, 3DM, 3DS, CGM. DAE, DGN, DWF, EPS, GIF, IGS, MTX, OBJ, PNG, SAT, JPG, SHX, STL, SVG, WMF

There are a few more, but I am sick of swapping between the program & this one - I keep losing my place! One of those has to be useful..... :confused:
There are still missing lines - I counted at least six or seven, but they look easy to fix, if I could just work out how. The snapping doesn't seem to work for me. I worked out the zoom though - There is adjustable scaling for the zoom, & yes, I can change the snap distance, but it doesn't want to join new lines to my existing ones. :sad:

I was going to put the latest file here as a DXF, but it's 10 meg before zipping, & I'm not going to start dumping stuff here hoping that someone can do something with it & take up all of Trossens server space. The latest file has some extra stuff now - This arvo I made a set of 21 longer fingers, so there is enough for all four hand pieces with one extra, but the knuckle, thumb & hand parts remain the same.

Slugman
10-27-2011, 10:36 PM
OK, time for an update.
I completed the rough/hand-made ankles & hips, then put them on Zychor to test. I tried to get him to stand on one leg, & the Bioloid foot twisted under his weight. I then grabbed some more scrap steel & made some strong & very heavy feet. The attached photo shows him on one leg. Keen-eyed Biped people might notice something interesting (Hint - Follow the LiPo wires from the battery) Quad or Hex people might not understand the significance of it.

So, must be time to get him to walk eh? Or maybe reduce the weight of the prototypes? Maybe even change the foot design so it doesn't look like he is wearing high heels when viewed from the back?
Um, no. I got distracted. :veryhappy:

The test parts came back from the Laser cutter, so I had a choice, make him useful, or waste time making him cool. :cool:
BTW - Thanks for your help elaughlin. I know it didn't work, but thanks anyway! I ended up having to re-draw everything from scratch, then the laser cutter guy fixed up my mistakes.

The 'Hand of Zychor' is 1.2mm thick stainless, & has 7 DoF, but I'm going to link together 2 fingers & both thumbs for curling them & moving them under the hand so I only need 4 micro servos to control one hand. That will allow me to have two hands controlled with the Arbotix if I want. It leaves me with 1 servo on each middle finger, for obvious reasons.
I have found a few design errors so far, & I will fix the final DWG before I post it up here eventually, but for those who cannot wait,

The support under the hand needs to be lengthened & bent so it's not in the way of the ball links on the side of the thumbs.
The guide holes on the back of the hand should have been 2mm, not 3mm so I could glue the cable shell to them.
All three fingers need lugs on the knuckles & corresponding holes in the hand to stop them rotating under pressure.
The inner wire guide for the thumbs are too long - Should have been shorter by about 2mm.
The distance from the first finger joint on each knuckle to the ball link controlling that finger/thumb should have been 2mm further away, so I would get more servo travel & therefore more torque for each finger.
I should have tried to make the joints 3mm instead of 2mm so I could have used nylock nuts. I'm gonna need a lot of thread-lock before I crush my enemies!
OK, so my enemies have to be less than 35mm high, but crush them I WILL!!!! :tongue:

The micro servos will be in a group on his body somewhere, either under or over the battery, & the cables will go along the arms to the hand. I haven't worked that part out yet. :happy:

Gertlex
10-28-2011, 01:16 PM
That's damn neat construction you're doing there! Got a front view of the bot, too?

(P.S. You can change the thread title under "Administrative" by selecting "Move thread")

Slugman
10-30-2011, 07:44 PM
Thanks Gertlex - I can't take much credit for how neat the hand is though - Laser cutting is really good, if a little expensive!
The other parts were roughly cut out, but I tried to get the locations of the pivot points fairly accurate. I never expected it to be so successful TBH. I thought I would have to make a few more versions before it would work. Those round brass parts on the AX-12s that the levers attach to are the hardest part to make, & the pics show the third version (4mm offset) which only need to be lightened & they will be complete.
My computer died on saturday while I was re-doing the DWG, so no front photos at the moment, but it's just a bioloid biped with an AXIS camera really. Hopefully the pooter just needs a new power supply.

Edit - While playing around with the hand after the pooter died, I found a few things - I cannot use 3mm bolts with the fingers as they are - Not enough clearance for the 0.036inch / 0.91mm diameter rods unless I make the fingers thicker/taller, so I will stick to the M2 bolts & lock-tite for future versions - I like the thin spidery finger look for a robot. It makes it more....Menacing I suppose.
Measuring the movement of the ball link on the knuckle, there is about 7mm of travel, while the link on the micro-servo has 10mm of travel, so I shouldn't need to modify them for more torque, & I have enough room to drill a new hole on the servo horn if I really want to. Otherwise, the only functional change not listed in my post above is the moving of those flanges at the side of the hand that guide the wires controlling the sideways thumb movement - They don't line up with the ball-links properly, which wasn't apparent on the sketchup model since I hadn't modelled the ball-links accurately.
I'm very embarrassed to say that I did later realise how to fix two of the three knuckles - I just incorporated the outer finger knuckles into the main hand, so that's two very rigid knuckles, & two less tiny bolts to muck about with. How I didn't see that while I was designing the thing I will never know. A real "Slap self on forehead" moment.

Slugman
11-04-2011, 07:04 AM
OK, here is my final .DWG of "The hand of Zychor". Not perfect, but it's the best I can do for now. I think I have fixed all double-lines & joined all the ends together. :happy:
Once I sort out the cables etc I might do a 'How to'. For now you will have to make do with the pictures earlier in this thread, but it's really not that hard.
It felt harder opening my wallet to pay for the laser cutting than it was to bend the hand into shape. :tongue:

Slugman
11-24-2011, 08:14 PM
Well, it's probably time for an update, although I took a backwards step with the hand. :sad:
It seems the cables are not just A, but THE problem. :genmad: I got the most bendy cables I could find, but they still have too much friction when bent through the back of the hand leading to the thumbs, so the 1kg servo just isn't powerful enough to overcome friction for one cable even with doubled torque, let alone two cables joined.
The tiny movement of the cable inside of, or stretching the sleeve is also a problem, even with the relatively short cables. It's not a case of the wire moving 5mm in at one end with 5mm out the other, but it sometimes moves 3mm out, sometimes 4, or maybe 5, depending on humidity, temperature, or phases of the moon. I don't know! :confused:
In short then, it looks like I have to go to a direct acting lever with the micro-servos to get decent & relatively accurate responses from the fingers & thumbs, with one servo on each movement. That's a total of seven servos if I keep the current configuration, & I like that config - It's at the robot end of uncanny valley, but humanoid enough to be described as a 'hand'. :cool:

It's back to the drawing board then, & I am sticking with my now tried & true method of prototyping in Sketchup, converting to a flat drawing, re-drawing in CAD, then finally getting it cut out by laser. Attached is a few photos, including the current sketchup WIP. There's one too many thumbs 'cos I need to test various positions for collisions & lever lengths. You can still see an old unmodified part colliding with a thumb servo under the hand, but the WIP shows how I am planning to fit the seven servos in the arm. An AX-12 will rotate the whole assembly as a wrist like the first hand.

Getting back to the original basis of this thread, those blue springs on the legs are very good for taking pressure off the knee servos when crouching - They are just shock-absorbers for RC off-road cars - They were nearly the perfect size to fit between the hip & ankle, & they bolted on with the Bioloid servo hinge used as a standoff. Very cheap, very easy, & very effective. One end could have used an M3 bolt to fit the hole in the end of the shocky, but an M2 fitted OK with a washer & the spring is always under some compression so it never moves. I like simple solutions that work, even though I wish I could get a nylock nut in the M2 size. They come with spacers to increase the pre-compression load on the springs, in case Zychor gets fatter, which is.......inevitable.

Slugman
02-05-2012, 08:45 PM
I've had the new hand pieces for five days now, so I better post an update.
The laser cutting look longer than expected, but these things happen.
So...... Design wise, it works. I just need to do a few minor tweaks here & there to the CAD to fix up some little stuff, & then I can get the second hand made. At the moment, it weighs 120 grams, so it's pretty heavy. I will be cutting some stainless out in the next revision, but since the servos are 7.5g each, there is only so much weight I can remove. Note that I have used 1.2mm stainless so far. so 0.9mm would be about 16g less.
I haven't connected it up to power yet & am going to test two different linkages to the fingers, which you can see in the pic - two heavy rods & one light spring-steel rod. I was going to use the springiness to give more compliance for the fingers, with rigid thumbs so I can grab whatever & not worry about over-straining servos, but it limits the strength of the grip. One link inside the hand is on there, although it needs to be lengthened 1.5mm, while the other one (Missing in pic) has to be re-done, as it bends around the other servo horn. One of the tweaks will be to allow straight links to be used.
It has taken me this long to post because I have been typing up the procedure on how to make it as I put it together. It's not complicated, but there is a definite order on how to make things fit with minimal screaming.
Anyway, enough talk. Ooooohhhh, a pretty picture! :happy:

Slugman
02-12-2012, 07:15 PM
Info on how to make the hand is now in the Tutorials section. :happy:

sarendt
10-17-2012, 02:30 PM
Slugman,

Can we see a video of Zychor?

Very cool work! Thank you for sharing!

Cheers,
scott

Slugman
10-22-2012, 11:15 PM
Sorry for not replying earlier, I have been away playing sport. We lost 9 games in 7 days. Sport was the winner. >:P
I got distracted from Zychor for a while due to real life & computer failure (The release of Skyrim didn't help either) but I got back into it about a month ago, & then made a tiny, tiny mistake. The wiring was all good, but I forgot to check the output voltage of the variable regulator that powers the Axis camera & both servo controllers, along with all the finger servos of course. They all wanted 5v, but that's not what they got. I haven't checked the hands yet - Been concentrating on the camera, which seems rather stuffed. I then went 1,000 miles away to play sport & drink stupid amounts of alcohol, & only just got back. At some point I will play around with the hands & see what might be salvageable, but when I flicked that fateful switch, only two finger servos actually moved, so I'm not hopeful. >:( I'll get there eventually.

Slugman
11-27-2012, 09:41 PM
Teaser photo time. :happy:
Decided to work on something that even I can't stuff up too badly. The pic shows an AX-12 modified so I can fit a small heat sink. It will only be used on the ankle & hip tilt servos for now, as the little aluminium part is a pain to make. It's 8mm wide, 15mm long, & the curve fits the motor which is 16mm in diameter. The heat sink is 21mm square, Part number HH8580 from Jaycar if anyone else down here wants to buy them. I just pulled the motor out, cut the rectangle out a fraction too small, then pressed the bit of ally into the servo case. The heat sink sticks to the outside nicely, & its black so it looks like it's designed from the factory to be there.
So does it work? No idea, as I only just finished the first one. It will obviously work to extend the servo use before thermal shutdown to some degree, I'm just not sure how much an effect it will have. No, I am not going to make a testing station to compare servos. I don't care that much. It doesn't weigh much either - Maybe a gram?!
If anyone wants to make one, there are two main options. I couldn't find a block of ally, so I used option 2.
1 - drill a 16mm hole out of a 15mm thick block of ally with the edge of the hole 2-3mm from one edge, then cut out the 8mm wide piece.
2 - get a flat length of ally 15mm wide & about 4mm thick, clamp it solidly sideways on a drill press so it rubs against a 16mm drill bit, then with a lump of wood push the flat side of the ally against the side of the bit so it mills out the curve, then cut out the 8mm wide piece.
I used thermal paste on the inside against the motor, & the heat sink comes with its own sticky pad which also transfers heat, so I didn't have to worry about trying to drill & tap tiny holes to bolt the parts together.
Once I get the other three servos modified to fit the ally parts & sinks, I just have to fix my home computer so I can play with PyPose again.

jwatte
11-28-2012, 12:08 AM
clamp it solidly sideways on a drill press so it rubs against a 16mm drill bit

Also known as "poor man's mill" :-)

If you have a vertical mill (or CNC) then I think there are two other options:
1) Cut 8mm strips from a 15mm wide, 4mm thick piece of aluminum bar. Clamp vertically in a vise, and mill sideways. You really need to get some backstop in the clamping for this to work.
2) Get a 4mm thick, 8mm wide piece of aluminum, as long as your work table is wide. Clamp horizontally. Get a 16mm ball nose mill. Mill out a channel in the middle. Cut to 15 mm lengths, as many as you can make out of the piece!

Let's ignore water jets, lasers, and die stamping, as I don't know of any hobby-level versions of those :-) (Although you can rent a waterjet for $2/minute runtime at a place I know in San Jose...)

I'm really interested in how this turns out for you! I've been wondering how the RX-24A could be cooled to take more load, and this problem seems very similar.