PDA

View Full Version : Full size robotic leg???



Jman26
12-02-2008, 04:43 AM
Hi All

I'm looking into the idea of building a sizable robotic leg with significant power to lift 55- 60 kg I'm an Aussie so that's about 130lbs. Does anyone have any experience with this? is it workable??? as the battery power is one thing.

What I was looking at for starters was the Dynamixle RX 64 (possibly the Ex 106 if required) for the hip and knee joint then the RX 28 for the ankle. I have checked the motor rating and they should be able to do the job. The prototype will only need 3 servos to start with, then also the CM 2 Controller and possibly an accelerometer and other sensors. The power will be provided by lithium ions possibly one or two 18.5V 21AH batteries from battery space they weigh 2 kgs (4.4lbs) each, and then a smaller battery for the controller. The frame itself will be either steel or carbon fiber depending on weight etc.

Any Ideas or suggestions??? I'm only a newbie to most of this so any help is appreciated.

Jarrod

DresnerRobotics
12-02-2008, 07:32 AM
Sounds like an ambitious project!

Now, just to put things into perspective: the EX-106 would meet your stated lift requirements if you kept its lever at 5cm, but you'd end up with quite the short leg. Anything longer than that and you would need something even more powerful. The ankle is a pretty high stress area as well, so a lower rated actuator would not suffice there. How much flexibility do you need with this leg? Most robot legs have a minimum of 5 degrees of freedom, only having 3 would limit your movement considerably. If its just something you need to lift perhaps look at a simpler design overall.

Adrenalynn
12-02-2008, 01:00 PM
Good post, Tyberius!

This is the realm of pneumatics and hydraulics, not hobby servos, I'm afraid. Welcome to the forum, BTW! :)

ScuD
12-02-2008, 02:42 PM
About a year and a half ago, I was working on this robotic lamp thingamajig, imagine a wall-E head on an arm descending from the ceiling. Or the fire-extinguisher robot from iron man, upside down.

Either way, I did some math on the design before plunging in on the build of the actual arm itself.

Came up with two 50cm pieces, an upper and lower arm, each with a motor/gear assembly.
More or less an elbow and shoulder actuator, really.

So here's the math:
The head weighed in at about 500gr.
That means the elbow motor needed a torque of at least 500gr*0.5m = 250gr/m or 250Ncm.

Next, take the upper arm. It has a weight of 500gr at 100cm, so that's 500*1m = 500Ncm, PLUS the weight of the lower arm and motor assembly, lets say another 500gr at 50 cm, hence 750Ncm.

That's 2.5Nm and 7.5Nm, respectively.

So I had two drillmotors with lots of torque, and figured it might just work. But then, you start thinking about how to get the motors attached to the arms, and have them geared up at a right angle.

I hobbed some simple worm gears, but then I forgot there's 7.5Nm of force pressing on the worm, so I'd be needing axial bearings. D'oh, another setback.

In the end, I gave up on this project due to the cost buildup and time it was taking me to get a 'simple setup' working.

Everything is possible, but you need to make sure you're in the right league.

Ps; if I can give you a hint, you may want to look into working with cable/pulley systems, it wouldn't work in my application (I care a lot about aesthetics in my projects) but it might be a solution to your problem.

jes1510
12-02-2008, 02:50 PM
One more note is that in the standard configuration with a servo connected directly to the knee then the thigh actuator has to move the weight of the knee plus the weight of the load with the moments added for each.

Adrenalynn
12-02-2008, 02:53 PM
And a note on Scud's, if I'm reading the design correctly - he'd have to add the weight of the drill motor at the elbow, too. ;)

ScuD
12-02-2008, 02:55 PM
Indeed, the location of the motor on the leg will affect the amount of torque needed, but either way the weight of the motor and the weight of the actual limb needs to be taken into account. (if I'm not mistaken, it's actually the integral of the weight, since the weight of the limb is at each point on the limb yet the torque arm increases... but that's too much math for me)

ScuD
12-02-2008, 02:57 PM
And a note on Scud's, if I'm reading the design correctly - he'd have to add the weight of the drill motor at the elbow, too. ;)

Ratz.. as I was typing it I thought about the combined weight of the upper and lower arm plus motor assembly, but screwed it up anyway.. I need sleep :happy:

Jman26
12-02-2008, 07:34 PM
Great thanks for the replys!

Yes 5 degrees of freedom would be great but i thought of starting as simple as I can. Flexibility isn't as important as the power, and I will have to go through the design of the frame motor placements thoroughly. The mesurements so far for the thigh section are 350mm from hip to knee axis and i'll get the mesurements for the shin section shortly.

I know it's a big project and ideally hydraulics would be the best but then you need the hydrulic pump and it starts to get to big for what I'm looking at, another thing is the response times hydraulics would probably not be as quick as the servos??? I'm at work now so cant type to much but I'll be back once I get home.

Thanks again

Rudolph
12-02-2008, 08:10 PM
Depends on how you build your hydraulic system, hydro can be pretty fast. Check out accumulators, it's a steel tank with a snazzy rubber bladder inside (like an inner-tube). The bladder is filled with nitrogen, the the hydraulic fluid is pumped into the steel tank, compressing the inner bladder. Then when you "hit the switch (valve)", the pressure stored in the accumulator moves the cylinder (or motor) faster than the pump can handle.

I guess it's not unlike a capacitor for hydraulics.

edit = more thoughts on the matter

Look into the hydraulic systems for cars too. You know those guys that hit the switch and launch the front end of their car 4 or 5 feet straight up in the air. They generally don't use accumulators, just gear pumps. They run like 20 car batteries to do it though. What kind of scale are you talking about here? (did I miss it?). Gear pumps are available in pretty small sizes.

Jman26
12-03-2008, 05:48 AM
Hi Rudolph

Will do. The scale would be quite large about 350 mm (13.79 inches) from the hip to the knee axis then about 315mm (12.411 inches) from the knee to the ankle. The frame I was thinking of building is in a rectangular shape that allowes placement of the components inside the frame.

I'll try to attach a pic of it.

DresnerRobotics
12-03-2008, 09:01 AM
Servos, at least the type that we deal with here (non-industrial actuators) will not suffice for a project of this scale. You'll have to look into hydraulics and/or industrial actuators.

ScuD
12-03-2008, 03:17 PM
Incidentally, I've wanted to try building some small scale (like R/C sized) hydraulics since I had my first lego technics set.

Does anyone know where I can find blueprints / exploded views of hydraulic components? specifically valves and gear pumps?

Might as well put the mill and lathe to work while it's cold out...

Rudolph
12-03-2008, 04:41 PM
It doesn't help right now, but I may have some drawings/schematics left in my toolbox. Next time I'm at my folks' house I'll check. I'll dig into my fluid power books and stuff when I'm there too. Probably this weekend.

Google images has a few exploded views of gear pumps right off the bat; http://images.google.com/images?hl=xx-pirate&q=gear.pump Not so much on the valves though.

Jman26
12-04-2008, 12:13 AM
Hi Tyberius,

would that be due to the length of the frame??? I did think about how much the torque would drop with that, and I'm not to sure.

Just say if we had the hip as point A and the Ankle as point B and an Ex 106 at the knee for power would that be putting out the same force at point A and point B as what is imput at the knee???

Thanks

DresnerRobotics
12-04-2008, 01:20 AM
It mostly has to do with the 5kg payload you had indicated. I've seen the eX-106 work beautifully on a robot where the distance between the ankle to knee, and knee to hip, was about 20cm or so, but this bot wasn't carrying 5kg on its back.

ScuD
12-04-2008, 02:18 AM
OT: Awesome, thanks Rudolph!

Jman26
12-05-2008, 11:29 PM
Hi guys another question..?

Do think the pololu Orangutang 168 or X2 would work with the Dynimixel EX 106??? Or does the Dynamixel CM 2 have the ability to be progammed? or do you have to turn the dial / switch with the CM 2 every time you want the motor to move? I dont think that's the case but just wanted to check.

Thanks

Adrenalynn
12-06-2008, 02:36 AM
I don't think the Orangutang will deliver more than 5v, will it? You'll be wanting, what, 18.5v? And it doesn't deliver RS-485, so you'd be adding a TTL->RS-485 shifter, and writing a library for it.

Of course, with your intended size and payload, it's just going to flop around and smoke the servos anyway - so I suppose it's ok... ;)

Bullit
12-06-2008, 07:43 AM
The Ex-106 would work well for this type project. The ideal configuration would be to have them in pairs set up as master slave which the Ex's are designed to do. This works really nicely because it provides a zero backlash solution and with 2 at each joint that will be like 300kg/cm at each joint at 24V.

DresnerRobotics
12-06-2008, 09:04 AM
The Ex-106 would work well for this type project. The ideal configuration would be to have them in pairs set up as master slave which the Ex's are designed to do. This works really nicely because it provides a zero backlash solution and with 2 at each joint that will be like 300kg/cm at each joint at 24V.

Wow, didn't know you could pair them like that. Mechanically, how would you set them up? Back to back?

billyzelsnack
12-06-2008, 11:35 AM
The Ex-106 would work well for this type project. The ideal configuration would be to have them in pairs set up as master slave which the Ex's are designed to do. This works really nicely because it provides a zero backlash solution and with 2 at each joint that will be like 300kg/cm at each joint at 24V.

$12,000 for just the legs!

I've been seeing some of the academic kids doing this dual servo setup and it seemed like a really bad idea as the servos can fight against each other. I guess Robotis gets around the problem with their extra synchronization cable. I wonder what it is actually doing...

Oh!! I think I just figured it out.. In dual mode it seems pretty obvious that the master servo would do all the feedback work for both servos, but... it could also only use just the master pot for both servos! That way you don't have to worry about constantly calibrating pots. Though it still seems a bit sketchy, but maybe it works fine in practice if you also monitor motor amps/temps as part of your feedback work.

Adrenalynn
12-06-2008, 02:45 PM
I bow to Bullit's _actual_ experience. He's the only one here who's actually done it, so if he says you can build a full-scale leg carrying 5KG with EX-106's - you can.

DresnerRobotics
12-06-2008, 10:08 PM
I bow to Bullit's _actual_ experience. He's the only one here who's actually done it, so if he says you can build a full-scale leg carrying 5KG with EX-106's - you can.

Well I thought we were talking full scale here as well, but it was later indicated that the between joint lengths would only be in the 31-35cm range, which changes things considerably.

Still very interesting to learn that the EX-106s can be used in tandem like that for double the torque.

Jman26
12-07-2008, 07:22 AM
I thought that might work but I'll need twice the battery power to run them for the same time.... so twice the weight...., but double Ex 106's should be able to do it. They are small enough to mount side by side in the frame. but i'll have to start saving coz I think that's about $2000 worth of motors.

I need one of those arc reactors out of Iron Man to power it ;-)

Thanks again for the help

Jman26
12-07-2008, 07:55 AM
Hi guys

I've done a quick drawing of this side by side set up. have a look see what you think

Adrenalynn
12-07-2008, 01:59 PM
So this is actually about a 2:3 scale leg, not a full-size leg?

I have long legs, but my legs are 57cm from ball of hip to ball of knee, and 51.5cm from ball of knee to ball of ankle. So 35/31 would be 66cm overall whereas my leg is 108.5 overall.

Jman26
12-07-2008, 05:45 PM
Yeah about that size so Dual 106's should be heaps (I hope)! or what about Rx 64's doubled??? that'd be a bit cheaper to!

DresnerRobotics
12-07-2008, 05:54 PM
So uhh, what are you planning on building here anyway?

sunithaya
12-07-2008, 08:38 PM
You have now quite a bit of replies that make a lot of sense. First thing I must ask though, why so much weight? can you explain a little more about the function, what is it that it is lifting (I imagine it is the leg, but again, why is it so heavy? is it already built for some purpose??? this simple question can help other folks give you more precise advice.

I also suggest you look into pneumatics (Adrenalynn mentioned that before) which can be extremely fast and powerful. Depending on the use and the management you could drive them with a pressurized tank and have a spare to replace while the other charges. (of course gas could give you more bang :eek: for cubic cm.

You mention in another post that you put a picture up? I can find it, can you point to it again?

cheers!

Rudolph
12-07-2008, 11:27 PM
OT: Sorry, ScuD, I didn't find much in my toolbox. Looks like I must have actually cleaned it out in the last few years. It also appears my fluid power books and such are in storage, at least I hope that is where they are...

Adrenalynn
12-07-2008, 11:32 PM
So just to clear it up - it is 2:3 scale or so, right, not "Full size" as the thread title suggests?

Jman26
12-08-2008, 02:56 AM
It would be a 2/3 size walking biped robot that I want to build for a uni project. Ideally with the ability to climb and decend stairs.
but i mite just scale it down to make it easier.

Here's the attachment of the pics again (this is just the thigh section)

Jman26
12-08-2008, 03:00 AM
Oh and the weight will mainly be from batteries, the frame and other sensors. (the armor and machine guns will come later) just joking

ScuD
12-08-2008, 03:49 AM
OT: Sorry, ScuD, I didn't find much in my toolbox. Looks like I must have actually cleaned it out in the last few years. It also appears my fluid power books and such are in storage, at least I hope that is where they are...

NP Rudolph, I searched google a while and found some really interesting things I'm gonna try. Thanks for looking!

Jman26
12-08-2008, 05:12 AM
Sorry guys I didn't say that other drawing was just of the thigh section here's a better overall pic of it. i'll keep you's posted on the progress. Thanks again for the help

DresnerRobotics
12-08-2008, 07:58 AM
My first and foremost thoughts on your design is that you're not going to be able to make it walk with only 3 degrees of freedom, at least not where you have them laid out. You need to have the ability to shift your COG between legs in order to lift the other. There is a reason that most humanoid robots have a minimum of 5dof per leg. The 2 and 3 DOF leg designs I have seen on smaller robots that do allow them to walk are nowhere near flexible enough to manipulate stairs (they also have immensly over-exxagerated walking gaits only possible because they are so small and light). You have take into consideration that getting biped robots to simple walk is a major challenge. Navigating stairs is something that even ASIMO has trouble with, and it has 6 DOF legs using high end actuators.