Anyone using linear actuators?

[Rblackmore]

Hi, Im new here and was wondering if anyone is using linear actuators such as solenoids or hydraulic actuators in their home built robos? Or are they hard to find in a small enough size? Thanks for any replies, just curious as my first thought would be to use a linear actuator instead of a solenoid.

[NewBreedWarrior]

Hey there...I dunno of what help I can be, but I have searched for linear actuators and have come up with little or no answer and that has been a while...I do not think that they could offer a small enough actuator to be fitted to a custom robo one bot..or for that matter if you are going in that direction...along with programing I have read that most LA's are harder to have control of using our boards..this is all I have read, maybe this is not as far as it could go..I am not any sort of genius, just letting you know what I found out...later

[Rblackmore]

Hmm, I kind of expected that to be the case =\
Servos just seem kind of slow/weak/bulky compared to linear actuators, but no need to reinvent the wheel

[gdubb2]

Hey Guys,
I was wondering about linear actuators myself. So far I have found this company in Canada that make some that look very promising. They are being handled by The Robotshop also in Canada. Links to follow.

http://www.firgelli.com/index.htm
http://www.robotshop.ca/

The product that seems to be the most promising is the "L" series, but it doesn't look like it is available yet. It would have a RC servo interface.

Pneumatics would be powerfull, but the cost in weight and complexity might be self defeating. Hydraulics would be out for the same reasons.

Good Luck, and please post if anyone finds more or better info.
Gary

[i-Bot]

It would be great to hear from someone who had used these.

Since it appears to be based on a geared motor feeding a leadscrew. If I apply no power, can I move the actuator ?

Do I need to maintain power to the motor to hold position when a load is applied ? Same question different way round.


I will ask Firgelli too.

[i-Bot]

Quick response on the question from Firgelli:

No you cannot move the actuator without power unless you get to about double the force capacity and back drive the gears etc

Maybe one could use an Open servo board over I2C for low cost control

[gdubb2]

I asked Firgelli about the interfacing of the things. The L-12 won't be available until early next year.

The feed back pot on the PQ-12 is 2K. I thought about just taking the board out of an old servo, but the feed back pot in a standard servo is 5K. Thats a little more difference than I like.

I asked about just on-off control, and driving the PQ-12 to the stops. They said it would cook the motor, so there isn't a slip clutch or limits in it.

[Rblackmore]

Pneumatics/hydraulics are so powerful and much quicker(atleast pneumatic actuators, havent seen very many quick hydraulic actuators =\) than servos, perhaps aluminum or such actuators would do. Those Firgelli actuators look good but kind of slow/weak as well. Shadow air muscles look great but their smallest(that Ive seen, I may be wrong) is 15cm fully extended so way too big for a robo-one bot. I wonder if it would be possible to make custom pneumatic cylinders? Eh, wishful thinking getting to me ><

[NewBreedWarrior]

o.O indeed a bright future for these, I can not wait till the l-12 and the pq-12 are offered. I do agree with maybe pulling some boards off of other servos. I believe it would be possible to have pos feed back along with vlt regs...we could do whatever we want as long as we have some motivation and determination to bring a new design of robo one bots or even greater than that...I will do some more research by the end of this week and pull some schematics of boards and voltages along with pos feed back.

[Joe]

My $0.02...

Solenoids are too weak to be of much use for moving joints. They might be strong enough to open a door panel in an R2D2-type robot, but that's about it.

Linear actuators based on a lead screw are of course much stronger, but they're also very slow. I doubt you could make a very competitive Robo-One bot with them.

I have some experience with pneumatics... these can be good, but it takes a LOT of power (in the form of compressed air, and something has to run that compressor!) to get much speed or strength. They also tend to be very fiddly, with springing leaks and things. Finally, it's surprisingly hard to find electrically activated valves that aren't hideously expensive (especially if you want to be able to do PWM with them to control position). Oh, and with any linear actuator, including pneumatics, you need to find some way to measure position if you want to have closed-loop control.

So, I think that the reason humanoid robots have pretty well settled on electric motors is that they simply provide the best combination of speed and strength for their weight.

However, I also think that standard servos, coming from an R/C vehicle background, are horribly designed for use in robotics. Even the companies now getting into robotics, like Hitec and Kondo, are still missing most of the features a sensible robotic actuator would have:

1. A through-shaft (instead of an output shaft on just one side)
2. Capable of continuous rotation without hacking
3. Simple two-way communication (via I2C or something similar)
4. Automatically shut down before self-destructing
5. Connectors that plug into the servos, so when you swap out a servo you don't have to redo all your wiring
6. Daisy-chainable to reduce wiring

Some new companies, like Robotis, are now Robotis, are now offering sensible servos like this, but the "old guard" companies are still behind the ball.

Anyway, if you want to avoid using commercial servos and design your own actuator, I'd suggest going for features like the above. No sense having to build a frame with a joint and then attach an actuator, when your actuator can simply be the joint.

Best,
— Joe

[gdubb2]

I looked at some slide action pots, there might be something there for feedback with the Firgelli. Use an external 5K with the actuator, forget about the 2K internal one. Then just use an old servo board.

I am looking at this for operation of grippers. For that I don't need blazing speed.

Gary

[Rblackmore]

Pneumatics or hydraulics seem best IMO but they arent common on such a small scale as robo-one. Servos seem to put out decent torque but I havent seen much faster than .13s/60* which is SLOW(again, IMO)

[RamDragon]

Been looking at the Shadow air muscle for the last half hour. It's a very simple design, and I think the "muscle" parts can be purchased at a local hardware store. The pneumatic connections would have to be machined, but that's relatively easy for anyone with the machines.

This really frustrates me because I just made a post yesterday stating that I don't believe pneumatics can provide a reasonable solution for robotics, being too bulky for small applications. The shortest pneumatic actuating rams I'm familiar with run about $30, operate at 60 to 300psi, and have an extended length of around 5-6 inches with between .75 and 1.5 inches of throw. They're made to move fast and cycle for long periods. The air muscle with 6mm diameter extends 150mm (about 6 inches) and operates under 60psi for a similar price.

The only concerns I have still is with the air supply and control. While it would be very easy to scale the muscle down to a very small size, small enough for robo-one style bots, the limit on the amount of onboard air would be incredibly restrictive. And, as Joe mentioned, pneumatic valves are ridiculously priced. I used to build paintball guns, so I've seen the extensive number of valves available and the corresponding prices. A low-end three-way valve (minimal operation for an electro-pneumatic valve) is around $100US. They get up to 12-way and custom-usage valves in the five-digit range quickly. And they ain’t small. There is simply no way to build a robo-one style bot with pneumatics. There's too much crap and not enough space.

So, while I still don't see pneumatics as a reasonable solution for robo-one, I can definitely see potential in it for larger robots. An Asimo-sized humanoid, for example, could benefit greatly in both speed and strength, from a combination of air muscles and servos.

I still have one question: Since my pneumatics experience is mostly through paintball, I find it difficult to visualize a system that can recover the air. I imagine that if a bladder were used as a reservoir that a system similar to hydraulics could be used, but then the speed would thoroughly depend on the speed of the pumps. If air is not recovered, then huge air tanks would be needed and they, in turn, would need frequent refilling. Am I missing something, or is that accurate?

Any way, those are my thoughts and opinions. Just one more question, since I couldn’t find and answer. Shadow’s website stated that they have positioning sensors for the hand; anyone know how that works? I can’t reason it out, but I’m sure it’s a simple answer.

[Joe]

I still have one question: Since my pneumatics experience is mostly through paintball, I find it difficult to visualize a system that can recover the air. I imagine that if a bladder were used as a reservoir that a system similar to hydraulics could be used, but then the speed would thoroughly depend on the speed of the pumps. If air is not recovered, then huge air tanks would be needed and they, in turn, would need frequent refilling. Am I missing something, or is that accurate?

I think the standard idiom is to exhaust the air to the atmosphere, but continuously top off your reservoir with an onboard pump. I've seen LEGO robots that work this way. In fact, the pump doesn't need to run continously — it's turned on and off as needed by a pressure sensor on the reservoir. (I actually kinda dig this idea for a humanoid robot, especially if the pump runs at varying speed depending on how empty the main tank is, since it means that a robot that's working hard would "breathe" hard, just like a human!)

Another issue you may or may not be aware of is that if the actuators are too far from the reservoir, they get sluggish (due to compression of air in those long feed tubes). So, most designs for good-sized pneumatic bots include smaller secondary tanks near or in the extremities, that provide a closer source of pressure for those actuators.

Best,
— Joe