Terminator

[sbates]

Howdy Dingo!

So I'm obviously not Australian, but I am thouroughly into robotics. I'm president of the local robotics club. I checked out the Terminator project, looks great. As you probably have already figured out, those $15 servos don't produce enough torque to be much use in anything large scale. Another draw back is that you tell the servo where you want it to go, but you get no real feedback to let you know if it actually moved to that position or what kind of pressure the "finger" is applying. Were you planning some sort of strain, pressure or position sensing system?

I am working on a little board to replace the servo brains inside the shell that can be give you various motion profiles and feedback on position, velocity, current draw, etc. I finally decided on I2C (TWI), but I really wanted to use the dallas 1-wire protocol so I could stick with the original 3 wire servo interface, but there are legal issues with making 1-wire slave devices. So my servo brain will be a 4 wire interface. Another nice feature should be the ability to chain them all together since I2C is an addressable protocol. That eliminates all of these "12 servo control" types of boards. They are nice, but still don't give the feedback I want.

I see that you are working with the air muscles. What are you using to control the air flow? When I looked into it, I found that I would be spending as much if not more on control valves than I would if I bought high-dollar servos ($40US each produce around 300zo-in).

Currently I am working out the same issues with hydraulics because I want to build something like the Parallax Hex walkers only big enough to ride on (about the size of a small car). The hydraulics are actually fairly cheep, but electronically controllable hydraulic valves are max$ and I need a bunch (6 legs, at least 3 valves each).

Do y'all over there have trouble getting parts? Do Digikey and Mouser deliver to you or do you have some good local parts outlets?

And, which way does the water swirl in your toilets when you flush?

Anyway, your site is cool and it looks like you have a number of neat projects going on. Keep up the good work!

--Steve

[Dingo]

Hi Steve!

Feel free to drop some Texan humour on us anytime!
No nasty moderators here


Actually the Terminator project is not mine but Rod's (username: Bones). He is the man behind such a great project. IF you want to ask him any direct questions you can just pm him here in this forum.

You idea of modifying cheap servos to include feedback is great! It would be a Godsend to all robot builders!. Now if we had pressure sensors on the ends too that would be awesome.


As for the air muscles,I know that Rod is also lamenting the fact the control valves cost so much. If I remember clearly, I think he was looking into machining them himself. I'm sure he'll post here to inform us all of what he intends.

Yeah, Digikey and Mouser *can* deliver to Australia but they want to charge huge amounts for postage. I'm not actually sure why, I send and receive stuff all the time to the States and postage doesn't have to be too bad.
Thus parts can be a bit of a pain to source easily. With Australia being about as large is continental US but with only 20 million people in it, there just isn't enough concentration of population in any one place for big electronics stores. Most of my parts are bought mail-order from Asia directly.

Oh and when you do get your giant hexapod built, Steve, make sure you get a mannequin, write "clever coder" on it and take a picture of the hexapod crushing the mannequin

I'm sure even the non-english speaking background users of AVR freaks will get it

Which way do the toilets flush? Um, downwards mostly ? Next time I have a kitchen sink full of water I'll make sure I note which way it drains just to make sure I'm still in the Southern Hemisphere

Anyway, welcome aboard Steve and just because your American we won't hold that against you here, (much ) just joking ;D

[Bones]

Hi Steve,
Thanks for the info. Those servos are for the fingers and wrist movements. I will be looking at position sensors for those movements and looking at pressure sensors for the hand.
I like the idea of your servos. Nice and clean control.
I'm machining up my own valve block and using servos to move the piston in and out for the exaust/hold/pump. There will be pressure sensors to measure the muscle's pressure.
The air muscles will be the same as Sonny from Irobot. They will be a bastard to program but look cool.
Have you seen the 6 legged tree log carrying robot thats on the net. It's like a loader. you get inside the thing and it walks around the forrest. I'll try and find the link.
Parts are ok to get over. Mainly I found you make your own if possible. In western australia it is one of the most isolated citys in the world and I find that they are not very innovative over here. I find i'm chasing all around town for things. Piss me off.
And the answer is counter-clockwise. Opposite way to you guys. Proberly the same as canada. lol ;D
If you have any ideas on the air muscles give us a yell.

Regards
Rod (Bones)

[sbates]

Dingo,

Feel free to drop some Texan humour on us anytime!

Thanks! I don't break into a rant too often, but some subjects just spark my flame. (like spammers in the other thread)

No nasty moderators here

probably just clever programmers... "They can't handle the truth!"

You idea of modifying cheap servos to include feedback is great! <snip> Now if we had pressure sensors on the ends too that would be awesome.

Actually I was thinking that current draw feedback is a direct measure of the amount of torque the motor is applying and a little math on the arm length or wheel diameter, etc. (maybe even inside the servo brain) is a direct measure of the "pressure" that the servo is applying to it's load. The concept is fairly sound, but various factors need to be taken into consideration. For example, if I have a heavy 6 inch arm attached to the servo horn and have it oriented parallel to the earth, the weight of the arm will cause the servo to "struggle" (draw current) to keep the arm in that position. However, if the arm is straight up, then the weight of the arm will not be a problem and there will be no current draw to keep it in that position. One can imagine all sorts of permutations of that scenario, but that's it in it's simplest form. I think some kinematics could take that stuff into account and calculate the actual pressure applied by the "arm" on it's load. Of course ideally we have a bunch of sensors, but in reality they cost $ and you actually find very few sensors in most robotic applications because of that. If there is some way to get the info without another sensor, then "good-on-ya-mate"!

Yeah, Digikey and Mouser *can* deliver to Australia but they want to charge huge amounts for postage. I'm not actually sure why, I send and receive stuff all the time to the States and postage doesn't have to be too bad.
Thus parts can be a bit of a pain to source easily.

Shipping prices are crazy these days anyway, even here (and Mouser is actually in Texas). I think it has something to do with crazy fuel prices.

With Australia being about as large is continental US but with only 20 million people in it, there just isn't enough concentration of population in any one place for big electronics stores. Most of my parts are bought mail-order from Asia directly.

I guess the grass is always greener on the other side of the fense. I'm sorrounded by millions of people all of the time. No matter where I turn there are hords of people. No matter which direction I shoot I hit a few. I'd love to move out inot a more rural area, but each time I try the hords just follow. On the other hand we have a number of electronics stores here in town. But they charge through the nose for single parts unless they are surplus. I generally don't use much surplus because I delude myself into believing that anything that I build could be sold on a web site and therefore needs to be reproducible. So I always design with real parts from reputable vendors and spend as much as possible. Someday I may learn that if I'm willing to spend a few minutes on it, scrap is generally what's left over after someone else did the reputable vendor thing and consists of parts that I can reproduce or still buy. So many projects, so little time...

Oh and when you do get your giant hexapod built, Steve, make sure you get a mannequin, write "clever coder" on it and take a picture of the hexapod crushing the mannequin

Now you're talkin'! "Death to clever coders!"

I'm sure even the non-english speaking background users of AVR freaks will get it

They made their position on the subject clear: "We support crappy coders and take offense to anyone who dares to call them what they are." I would not want to offend anyone

... toilets flush? Um, downwards mostly ? Next time I have a kitchen sink full of water I'll make sure I note which way it drains just to make sure I'm still in the Southern Hemisphere

I learned all of that stuff in school, but like most things in science, it makes sense, so we just accept it at face value. Since I have a direct line to the other side of the planet here, I thought I'd check the facts just to make sure it was not a load of hooey.

Anyway, welcome aboard Steve and just because your American we won't hold that against you here, (much) just joking

Well thanks, and I won't hold your being an Aussie against you either. I'll even try to refrain form picturing you all as "Crocodile Dundee" and his sidekicks or Steve Irwin: "Crikey, He's a big one!", if you will refrain from picturing me as "Walker, Texas Ranger" or the folks from "Dallas" or some old John Wayne western. Contrary to the beliefs of most of the rest of the planet, not all Americans feel the need to get together with a couple hundred thousand of our best buddies and about $50 billion in hardware and tromp around the world "liberating" all of the oppressed peoples.

--Steve

[sbates]

Bones said:

Thanks for the info. Those servos are for the fingers and wrist movements. I will be looking at position sensors for those movements and looking at pressure sensors for the hand.

Do you have some ideas for the pressure sensors?

Bones said:

I like the idea of your servos. Nice and clean control.

Thanks! I finished up the design last night. It is based on an L293 and an ATtiny26L. The board is less than 1 inch square and only double sided. Should fit inside the servo shell in place of the existing controller nicely. I'm gonna try out pcbfabexpress.com and should have the boards back by end of next week or early the week after. Meanwhile I'm working on code for the AVR. I'm gonna try to use the GCC C compiler and see how that works out. I'll let you guys know how it goes...

Bones said:

I'm machining up my own valve block and using servos to move the piston in and out for the exaust/hold/pump. There will be pressure sensors to measure the muscle's pressure.

You're gonna need some strong servos ($$$). I'm planning to use standard hand control valves (cheap) for my hydraulic applications, but I don't think I can afford servos to drive them all, so I'm gonna try DC motors with lead screws, but then I still need to add some position feedback. Fortunately, I should be able to use my servo brain to control them. In the end the whole apparatus may cost me more and be less reliable than if I'd just bought commercial electronically controllable valves.

Bones said:

The air muscles will be the same as Sonny from Irobot. They will be a bastard to program but look cool.

In robotics, looking cool is the best part. Are the air muscles very springy? They look so cool, but I have not had a chance to actually play with them yet.

Bones said:

Have you seen the 6 legged tree log carrying robot thats on the net. It's like a loader. you get inside the thing and it walks around the forrest. I'll try and find the link.

I have heard rumors, but I've never seen one. I would REALLY like a link if you can find one. Most of my friends and family just roll their eyes and shake their heads when I begin talking about my project ideas. It's good to know that I am not alone.

Bones said:

Parts are ok to get over. Mainly I found you make your own if possible. In western australia it is one of the most isolated citys in the world and I find that they are not very innovative over here. I find i'm chasing all around town for things. Piss me off.

At least you can get around town. It's so crowded here, that it's virtual grid-lock all the time. Almost impossible to move around. There has got to be a happy medium somewhere.

Bones said:

And the answer is counter-clockwise. Opposite way to you guys. Proberly the same as canada. lol ;D

Ok, there is something wrong here. I just flushed two different toilets, both went counter-clockwise. I figured that it was just the effect of the fill "jets" so I filled two different sinks with water, let it stand for a moment to "settle" then pulled the plug. Again both went counter clockwise. It could be the effect of Oklahoma: "You know what keeps Texas from falling off into the Gulf of Mexico? ... It's because Oklahoma sucks so much!". ... or I could blame the Canadians, guys are bass-ackwards about everything!

Bones said:

If you have any ideas on the air muscles give us a yell.

You bet. I have a nice little compressor and portable air storage tank. Maybe I can find some time to play around with them. What kind of mesh are you using around the rubber hose? Is it that nylon stuff? I might have some of that handy around here...

--Steve

[Bones]

Hi Steve,
Here's that web site for the forrest walking Vehicle.
www.plustech.fi/Plmain01.html
As for my designs. The valve unit and pressure sensors are still on the design bench but the valve design should be ok with basic servos.
They won't need much pressure to turn on or off. The other I was thinking of was fuel injectors and a valve design using muscle wire.
Because of the air flow through them it will cool the wire quickly for closing. I have seen some out there that fit in a large fuse socket that look alright.
I'll reem out a block of alloy and make valve block to see how they go. A mate has already made some and he said they work ok.
Any info on those servos and ideas would be great.
I will be putting full design spec for my robot up on the site so other people can copy it as I go and perfect each section.
I think it might take a while but hey learning heaps.
That mesh that I use is for keeping cables together. I got it from and electrical cable supplier. Have a look at the following sites.
brl.ee.washington.edu/Research_Past/Biologically_Based/Device_01_McKibben/Mckibben_Costruction.html
members.rogers.com/cybotix/cybot/muscle.html
www.imagesco.com/catalog/airmuscle/AirMuscle.html
www.techflex.biz/

Rod

[sbates]

Bones,

Thanks for the link to the forrest walking John Deere! It is similar to what I want to build. The main difference being that it's legs are presumably designed to make it as narrow as possible for the forrest environment. I want to do something a little more like a spider or the parallax hexpod (www.parallax.com/html_pages/robotics/hexcrawler.asp) and of course mine will look like a go-kart compared to that commercial rig. We'll see. I've been seeing a bunch of these mini backhoes (www.ditchwitch.com/dwcom/Product/ProductView/10556) around here and my first thought was that maybe I could just buy six of the arm assemblies and build my own body to attach them to. But the price is out of this world, so I have decided to make my own. The metal and hydraulic actuators are not too bad, but the valves are where the money is.

Those are some cool links on the air muscles. My only concern about air muscles in general would be their "springiness". I like the idea of using airflow to cool the nitnol and get a quicker reaction. Muscle wire is cool, but power hungry and fairly slow on the cool-off side. So using it for the valves and air flow cooling are great ways to overcome those issues. I'll keep my eyes open for your updates on those valves and update you all with servo info. I'm looking forward to hearing about your experience with the air muscles.

Learning is the fun of this hobby. It's clear that others have already done 6 leg walking robots and air muscles, we are not breaking new ground here, but it's the learning and overcoming the technical challenges that make it fun.

--Steve

[Bones]

Hi,
Glad you liked the link. One of those back hoes with spider legs would be great. I know a professor over here in WA.
He said he built some robot spiders while doing his UNI in Sydney. It was at one of the museums over there. He said it was on an conveyor belt and when you pushed the button on the display case the belt would start. The spider detected it moving and would start walking and then running.
He was telling me that he had them running up and down the hallways at UNI.
Imagine someone coming out of a door and one of these spiders running at them.
After I finish this humanoid robot he I'll be doing a spider robot. May one for Robot wars that can run and jump and pick apart the other robot(I mean remote control cars with blades on them).
As for the air muscles and the springiness, Thats what I'm looking for. If it is too riget then you start breaking things. You need a bit of play or give to cusson stuff.
The air muscle works well in that way.

Rod

[sbates]

Bones said:

May one for Robot wars that can run and jump and pick apart the other robot(I mean remote control cars with blades on them).

Rod,

Thank You! I thought I was the only one who realized that it's not "Robot Wars" it's "RC Car Wars". In fact, I was watching TV one fine Sunday afternoon when "Robot Wars" came on. I finally had enough and said out loud: "What a load of crap! Those are not robots! I'll show 'em what robots are!" I immediately went to the web find a local robotics club to join and found none, so I started one. My original goal was to host an event similar to Robot Wars, only with real robots. The only remote allowed is a start/stop. My second goal was to move it outdoors or at least onto some kind of rough surface. If you watch RW a few times, you can see that the simple low ground clearance wedge with pnumatic flipper is a real winner brcause of the smooth floor. Get rid of the smooth floor and the remote and now you've got a REAL Robot War. Find, Seek, Destroy!

--Steve

[Bones]

Sorry no reply for a while. Christmas and I work in retail.
But I have made my shoulder joint for my robot. I might as well get the hard stuff out of the way first.
If I could get you to post them for me please Dingo. I email them to you. When you get time.
Oh yeh Steve. Nothing worse than a remote control mower being called a robot.
I think your idea of getting out in the rough and fighting is better. I was thinking that you should only see what the robot see's. Therefore more fun. Less risk to people and more weapons.
Design plans are still coming for my robot. Bit by Bit

Bones

[Dingo]

Yep I'm back from the great Xmas black hole now

I'll post the pics asap

Oh yeah I think seeing just what the robot sees would be cool too. Plus some action replays from outside when one autonomous bot wastes another.

[Kiwi]

Hey!

Im a mechatronics enginnering student studying at Auckland, New Zealand...

This place seems to be pretty full of good ideas and expertise, so I thought id make the odd appearance.

My forth year project is to design and build a robot hand, which is quite fun but also pretty challenging as since its uni, they want you to work out all the kinematics and control...

Anyway

Do you have some ideas for the pressure sensors?

We are using piezo strips, which are apparently quite cheap, they create a voltage when deformed, so sandwedge one between 2 pieces of rubber or something, when that rubber is pushed down a voltage is induced that can then be mesured or turned into a digital signal or whatever you want...Well thats the idea anyway, we are still experimenting with it but Ill let you know haow we go if you want.

Also an incredibly cheap way to get rotational position sensors is to buy an old mouse (you can get them for about $10) and steel both of its X,Y axis encoders.

[Dingo]

Welcome abord Kiwi!

That is a good idea, with the rubber on the outside and the sensor inside protected.

Would love to see some photos or video once you get the hand together.

[Bones]

Hi Kiwi,
Sound like you got a few ideas. Got any info on the hand project you did. I'm starting to mill the finger soon and am wondering how you did yours.

Bones

[Kiwi]

Well unfortunatly for us, we have to design for manufacture, so it is easy for the workshop to machine but wont look all to beautiful but will be functional.

At the moment we are getting one finger prototyped. I designed the parts to be cut from aluminium U sections.

We chose an underactuated design which basically means you get 3 degrees of fredom with one actuator. Check out: wwwrobot.gmc.ulaval.ca/recherche/theme04_a.html, for the general concept and some interesting videos.
Hmm ill try and post a jpeg of the Pro E model.