Complicated Project (Hydraulics & More)

It's called, "robotics", a 10 megahit Google term.

DAGS on "air muscles". They're marketed by a British company. While you're on the site, check out how their built their robotic hand -- and watch the video of it picking up a pint.

There's a lot of good information there on how they built their hand, including stuff on power sources.

--RC

Some first questions: Just how large is this hand? How strong is it's grip to be? How sensitive it's touch (should it pick up an egg without cracking it)?

Precision control with hydraulics has been done for half a century, on aircraft. If you have access to a military surplus place, you might be able to pick up servo valves, which will take a small electric current input, & convert to a flow of oil. Now add some kind of positional sensor to the motion, and its a straightforward control problem.

As to power supplies, hydraulic power (oil under pressure) is supplied by a pump. These come in all sizes, and are driven by any available source of mechanical power (electric motor, diesel engine, whatever). The equivalent of power reservoir capacitors is an "accumulator", which is a tank containing (usually) nitrogen under pressure. Oil pumps into the tank, compressing the nitrogen further. The oil can then be drawn off as needed. Remember that most hydraulic pumps shift a constant volume of oil per revolution: driven at a constant speed, they are like an electrical constant-current source. You *must* provide a pressure relief valve (ie a shunt regulator) to allow the oil to escape when your cylinders don't consume it. Else, something will blow...

The power required is a simple matter of force * speed. How much force shall the biggest part of your hand exert, & how fast can it move, while exerting that force? (Then double it - at least - when sizing your pump).

Do remember that hydraulic oil, under some 2000psi pressure (typical) will escape through any less-than-perfect joints. The fine spray produced will pass straight through your skin, & given you a very nasty case of blood poisoning. As always, safety first...

Pneumatics are very hard to use when you want precision of movement. This is because air is compressible: it's like trying to position something accurately, when you have to push it through a large spring. Very difficult.

If you're new to the mechanical side, how about building a small-scale model first, to get the feel of things? Stepper motors driving leadscrews could be a very practical approach here.

"Jake" wrote:

:I'm posting this to this group because it seems that this is where most :of the hydraulics related questions go. : :I am planning to start a rather ambitious project that doesn't seem to :have been attempted very much so far, as far as I can tell. : :One of the largest obstacles to completing this project is that I :really have no experience in metalworking. I certainly don't know :anything about hydraulics but what I've learned on the internet. I :know very little of electronics, or at least the kind that will :probably help in this project. I am a computer guy first and foremost, :and I can write programs in many languages, but I just haven't done :much work creating machines. : :So here is what the first step of my project is: I want to create a :large mechanical hand. I also want to create a "device" or "interface" :to control said hand. : :I have the general concept worked out in my head - I haven't actually :drawn up a blueprint or schematic yet. The main reason I'm only at :this point is that I don't really know a lot about hydraulics. I know :that: :Hydraulics are strong, that's why they are used in construction. :They are expensive. :They can be dangerous if not treated well. :And that's pretty much all I know. : :What I'm hoping for from you guys, is some input on the following: :How small can hydraulic cylinders be? I'm shooting for something like :a 5:1 scale. Would it be feasable or possible to put one cylinder on :each "knuckle" to act as a muscle for that part of the finger? Or :should I be looking at having a kind of "tendon" control system where :each finger is controlled by a series of cables that are in turn pulled :on by one hydraulic cylinder per finger? Bluntly, are there hydraulic :cylinders that are less than a foot long and only about 1 or 2 inches :in diameter? How much power would one of those generate? : :Should I be considering something besides hydraulics? Such as electric :step motors or pneumatics? If so, what would the pros/cons of going :with one of these alternatives be? : :As far as controling the hydraulic cylinders - is there a way to tell :the cylinder to only go out a certain distance or to only come in a :certain distance? The only control system I've ever seen for a :hydraulic cylinder is basically a simple extend/retract/stop system, :that has to be attended to by the operator (i.e. I want to say :"Cylinder: Extend 3 inches and stop." and then have that happen). : :What is the maintenance like on hydraulic systems? What is the most :often replaced piece of a h.system? : :What do you do to power a hydraulic system? Do you have to lug around :a huge battery or do you somehow convert energy from an engine into :hydraulic power? : :Once more on controlling these things - do you know of any way to hook :a hydraulic system up to a computer and control it via the computer? :Like if I were to create a program that could talk to the hydraulic :system through my computer then I could just control the whole thing :from my computer screen. : :I think that is enough for now, that's an awful lot of questions! :Please help me if you can. If you know of another place I might ask :these questions please feel free to let me know, the more information I :can get on these topics, the better.

Using sensors and feedback make pneumatics a whole lot more accurate. Given that OP is apparently trained as a programmer, I'd think this would be a good way to go.

See:

for the air muscles and for a little on sensors and feedback with pneumatics in robotic applications.

--RC

"Jake" wrote in news:1108369621.374189.304230 @f14g2000cwb.googlegroups.com:

Force = surface area of piston x pressure. You need to subtract the rod area from the piston area on the rod end to get the real surface area.

Yes, hydraulic cylinders can be fairly small indeed. But I really do not recommend them for this application.

Yes. The cheapest method is probably going to be stepper/servo control and actuation.

Yes there is, but it is not easy, nor cheap. You will need a proportioning valve for each cylinder you wish to control, you will also need a feedback device for each cylinder (encoder), and you will need control electronics (a drive). Proportioning valves alone will be in the several hundred dollars _each_ range.

A pneumatic system is basically the same in principle of hydraulics.

Hydraulic systems are somewhat maintenance intensive, if properly maintained. If you are using proportioning valves, maintenance becomes critical, as even tiny contamination is your dire enemy. Then consider all the hoses you have, that are going to be moving within this arm, being abraded. Having a 3000 psi hydraulic hose spring a leak of 120° oil is not a fun thing if you are nearby.

Well..you might want to bring several batteries to drive a hydraulic pump, depending on the pressures/force you need available.

Yes, we have many machines which do this, but as I stated, it is _not_ cheap.

Pneumatics can be easier and cheaper to deal with in some circumstances. It's certainly a lot less messy when you spring a leak!

Get to know your local industrial surplus sources. Lots of pneumatic and hydraulic surplus is available, but you have to know where to look. If you're in southern california, one retail-friendly surplus house is C&H Surplus in Pasadena.

Hello Jake, Try this site.

On 14 Feb 2005 06:37:04 -0800, the inscrutable snipped-for-privacy@btopenworld.com spake:

The page cannot be found The page you are looking for might have been removed, had its name changed, or is temporarily unavailable.

Please try the following:

• If you typed the page address in the Address bar, make sure that it is spelled correctly. * Open the
  formatting link
  home page, and then look for links to the information you want. * Click the Back button to try another link.

HTTP 404 - File not found Internet Information Services

---------------------------------- VIRTUE...is its own punishment

Website Applications ==================================================

spring.

But if you want precision force, rather than precision positioning, then a natural spring of the right force constant is a good thing.

Imagine trying to pick up an egg with a robotic hand when you have hydraulic positioning - very accurate but no automatic control of applied force. Then imagine it when you've got low-precision positioning but good control over the force applied.

What you really need for a robotic hand is the "right" mix of position accuracy with adjustable force. You won't always be picking up eggs!

Tim.

All these questions and you forgot to say what you wanted the hand to do.

Oddly, no one else has asked this question before me. By asking non-specific questions, you either get the shotgun effect of everyone posting little tid-bits that may or may not be helpful, or you don't get very many (any) answers at all.

So:

1. What do you want the hand to do? This is _the_ most important question to answer, the first question to answer, and the ONLY question that you absolutely must answer _completely_ before you can go any further.

Once you've got the answer to #1:

1. How much money do you have?

1. How much time do you have?

2. Do you personally know people who can design and machine/fabricate mechanical parts, install hydraulic fittings/valves/pumps/actuators, design and build control circuits, etc., etc.?

The above questions have to do with "feasibility". There will be other questions, but start above.

Regards,

Robin

You need to define what you want your project to do before you decide (or ask others) how to do it. You need to know what forces, speeds, and precisions it should have, and what budget and capabilities you have or can obtain or develop.

Example: hydraulics are good where force and motion is needed from a relatively small actuator but nowhere is it written in stone that they must run at 3000 PSI, use oil, or even be made of metal. Nearly all industrial hydraulics do use metal, oil and high pressure to get the very high forces usually associated with hydraulics, but you may not need your "hand" to be able to crush rocks. If ounces of force would suffice, then small electric motors and leadscrews might work well.

I've seen all-plastic hydraulic devices that were all plastic and used water as fluid: remote throttle actuators in a dynomometer cell where the operator simply wanted to be outside the region of loud noise.

you should go to a toy store and buy a Lego Tecnics set. Pneumatic actuators, lots of connectors, very cheap...

I'm going to lead off by saying I'm very pleased by the responses you have all given - they have been helpful in getting me off to a good start. I like the idea of the air muscles mentioned earlier. They certainly look promising. But you have raised some very important issues, Robin S. Here are some more details:

I'm probably not interested in spending any more than $2000USD on the hand. It would be nice if it could crush rocks but not necessary. As far as size and strength goes you might imagine one of these hands attached to the arm of one of those big pieces of construction equipment. I think the general idea is to get a hand working, and then build an arm to attach it to, and then a shoulder, working all the way until I have a giant ... "robot"?

I'm very much intrigued by the idea of an exoskeleton, only I think it would be quite difficult to make one that is close to the size a human

- at least for a person working on it in their garage without some sort of big financial backing. I think it would be easier from a design/engineering standpoint to build a large-scale exoskeleton, like a large humanoid vehicle. That is why I want to build this "hand". It seems like a good starting point, I will have to work out the actual construction of said hand as well as the system used to control it. As much as possible I would like the operators hand movements to be translated to the mechanical hand for the most part 1 to 1. So if the operator does a thumbs up, the mechanical hand would do one too, in as close to realtime as possible, meaning that the mechanical hand forms the thumbs up as the operator forms the thumbs up and they both stop moving at approximately the same time.

Some sort of force-feedback mechanism would come in handy here, because it would allow the operator to control the gripping strength of the mechanical hand just by interpreting the feedback from whatever pressure/force sensors are installed on the hand. What may be required is some sort of computer layer in between the operator and the hand, so as to allow for the gripping of squishy or delicate objects like eggs or people (Imagine a big hand pulling off the wall of a tall burning building and extricating the people who may be unconscious inside) without crushing the held object. I doubt the size of the mechanical hand I have in mind would allow for the gripping of an object the size of an egg. Perhaps the smallest object it could grasp would be a softball or a mushball (about 6" in diameter, like a cabbage).

At any rate... I figure before I go off to build a 20 foot walking mechanical robot suit thing I should start small - thus, the hand. At the very least if I were to be able to build such a hand there might be some market for it in construction or other areas if it were to advance enough to really be useful.

How strong is the human hand? I envision my mechanical hand to be on the order of 5 to 20 times stronger than that. Probably close to ten times as strong. But I don't have any specific application for it as of yet so it is hard to say. I just want to build one. I want it to be articulate, and I want to be able to control it with my own hand.

I can't remember the name but someone mentioned Legos and the Technic pneumatic pieces - that's a stunning idea. I'll try that out as far as prototyping goes to get the structure/design/frame figured out.

A couple of questions that just came to mind - or just one:

If you have two identical hydraulic cyclinders next to each other, operating on the same object and working off of the same base, is the net force just the sum of the forces of the two cylinders? In other words, if I have a platform that is lifted by 4 hydraulic cylinders attached at each corner, and each cylinder can exert a force of 25 newtons, is the total force 25 newtons or 100 newtons? I think it would be 100 newtons.

Ok I'll let you guys comment now on how insane I must be to dream of making something like this... :)

[...]

You can make your cylinders basically any size you like--the force generated equals the oil pressure times the area of the piston, so you can scale 'em down a bunch if you're willing to deal with higher pressures. And with a "hand" you probably won't be needing a whole buttload of force on the intricate bits like fingers--you can easily build a system using different pressures in different sections. A piston less than 1/4" across has an area of 1/2 a square inch, so

20psi oil pressure gets you ten pounds of force, which ought to be plenty for a finger. Step it up to 50psi for hand-smashing fun and you're still dealing with negligible pressures and won't need hydraulic-use rated hardware. Safe to work with, too. Generally, as components shrink they can handle higher pressures more easily, too, since there will be less surface area internally. (for example, a hypo needle has a burst pressure in the neighborhood of 10,000psi) If you don't need a whole lot of flow use small lines. Cheaper fittings to boot. Size it so fluid won't have to move faster than 20 feet per second. If you use hydraulics your biggest headache in terms of construction would be the lines and hoses--that can be mitigated somewhat if you place all of your valves as close as possible to the cylinders, so you might prefer the tendons option. Also, if you go with a tendon setup you have more room to install positional sensors either out at the joints or along the tendons themselves. And with a tendon setup the cylinders won't have to move around, so you could plumb them all with rigid lines, which are better/cheaper than hoses to work with. I assume you've read about open center and closed center hydraulic systems. Open center basically has a low-pressure standby mode where oil just gets pumped straight through the valves and back to the tank. It's easy to do if all of your valves will be located on the same manifold, but gets quite a bit more complex if you don't. Closed center would be the way to go in that case, but you'll need either a variable displacement pump or some sort of unloader valve arrangement. With a variable displacement pump you'll wind up with a rather high pressure standby that could make the hand difficult to control. If the valves will all be computer-controlled and not in a common manifold, closed center with an unloader activated by the computer would probably be best. Make sure to always have a pressure relief valve in case something goes wrong. Maintenance depends on how cleanly you assemble the system and how hot it'll run. You'll probably be running pretty cool, so it's not too likely that you'll wear out or burn the fluid any time soon. Make sure your reservoir has a low spot in it and the pump's intake is higher than it. That way any water that might get in will stay out of the hydraulics. You'll get mixed opinions on where to place the filter--some say on the return to the tank, some say before the pump, some say after the pump if working pressures allow it. On the return to the tank is a logical choice if you know you can put the system together cleanly. Pump intake is a somewhat dangerous place because if your filter gets clogged too badly it could either cavitate the pump, or worse, get sucked into the pump. You could also use a bypass filter setup--a secondary high flow, low pressure pump that does nothing but take fluid from the reservoir, pump it through a filter, and back. Tie the main pump's intake to the filter's output and you'll get the best of all systems and a low-pressure filter. If you decide to build parts yourself you can make hydraulic cylinders pretty efficiently with DOM (drawn over mandrel) tubing. You only need to clean up and thread the ends and install your plug and gland nut. Of course, you'll need to make plugs, pistons, and gland nuts. (: As far as hydraulic positioning systems, I know some hydraulic cylinders that need accurate positional feedback have a ballscrew that is attached to a rotary converter in the base of the cylinder. The nut is on the end of the rod and the rod is hollow to make room for the screw. But I think something like that would be extremely expensive and not available in smaller sizes. Also, some cylinders actually have threaded rods that are then built back up. The rod is smooth when all is said and done, but you can place a magnetic pickup by the rod and count as the threads pass by below the surface. Again, expensive, but I think more readily available in small sizes. You'd probably be better off using a sensor outside the cylinder or even at the joint.

I would recommend you go in this direction before you invest some very siginificant funds into something that could take years of work to get going.

The only thing that matters is the area of the surface which is perpendicular to your force and on which hydraulic (or pneumatic) pressure is acting upon. In the case of a group of hydraulic cylinders, this is the total area of all the pistons.

You should be aware of the forces you are (possibly) dealing with here.

Lets say a system pressure of 1500psi (~100 bar), a cylinder with a piston diameter of 1" (~25mm).

F=pi*r^2*P

Where: F is force in pounds pi is pi (~3.14159...) r is radius of pison (circle) in inches P is pressure in pounds per square inch

F=pi*.5^2*1500 =1178lbs (Which turns into roughly 5200N)

Obviously, this is a _lot_ of force. While it's fun to see rocks get crushed, you may not personally enjoy the feeling. Hydraulic systems are not designed to fail when they get bloody, so you really only get one chance (and typing code without a full compliment of fingers must suck).

A guy at work recently got caught in an automatic surface grinder's hydraulic feed mechanism. It nearly sheared his finger off.

Well, at least you see what we're seeing. You should definately see what you can do with this Lego Tecnics set (if it even exists). Anything that speeds up the prototyping stage of development is going to be a _huge_ help.

You've got a lot on your plate...

Regards,

Robin

The "snakes" in the movie _Anaconda_ had Hundreds of hydraulic cylinders at every joint. All controlled by a computer program. Th central hydraulic pump was pretty impressively large too. Programmers worked all night to program the scene for the next day. Pete

No, that would be 5/100 sq" (.049) for a 1/4" dia. piston (extend, only. The rod would reduce the retract area and force.)

So, for a 1/4" piston, you get just under ONE pound for 20psi.

LLoyd

Somewhere I have a letter from the DOE/NIST stating that water pressure can not work for hydraulic devices. And furthermore if water pressure is lost the solar device will burn down cause the sun will still be tracked.

On Tue, 15 Feb 2005 21:40:02 -0800, Sunworshipper

All generalities are false....