Robotic joints construction

The squarish things protruding from most of the joints on those robots
are servo motors. Some may be attached internally to gears, while others
may directly move the joint.

This is readily possible given the scale of these robots. The picture
makes them look small, but they are actually quite large.

The concepts are the same for smaller robots. Look at the R/C
servo-driven humanoid robots like the Robo-One or the Robonova for ideas
and inspiration.

-- Gordon

"Gordon McComb"

On a similar note, a couple weeks ago I took my family to legoland and I was
able to ride their "knight's tournment" on level 5. They are very similar to
puma arms, and you have a good opportunity to feel their strenght as you're
literally shaken in all possible manners.

Cheers

Padu

RoboOne:
  http://robogames.net/photos/robo-one01.jpg
Robonova:
  http://www.team-iwan.de/bilder/robotliga2005/P1000634k.JPG
(price around 800 USD for both?)

Seems they all use _direct_drive_ from servos:
  http://en.wikipedia.org/wiki/Servomechanism

Which leaves me with two questions. How strong are these r/c servos really?
And how does the controller device (computer) know where the limit of the
feedback potentiometer is?, as variable resistors tend to change in value
depending on temperature, age and usage. I would prefered optical feedback
but that's just me I guess =)

Is it feasable to make an "arm" to place something with a .3175 mm precision
with these circumstances .. ?

An idea is to use a rotateing prism from a laserprinter and led. To get
feedback and achive precision that way provided the servos have enough
stability.

Cheaper that way, but it can be done using indirect means as well,
including belts and chains.


"Strong" is relative. Strong enough to make a 1-foot high walking robot
but not strong enough to pick up a six-pack. Once you define "strong"
you can determine if there are any off-the-shelf servos that provide the
torque you need. Given the latest digital models, which have torques up
to and exceeding 150 oz-in, you can usually find what you want, assuming
a smallish robot.


You can always add optical feedback for indexing purposes. However,
except for the digital models the deadband of most servos is about 6-8
microseconds, meaning that there is a lower limit to resolution.
Assuming 180 rotation, and a (typical) 1200 microsecond range for that
arc, you're looking at 6.66 us per degree. That's about within the usual
deadband specs of analog servos, but it also means you have no better
than one degree resolution. If you need something more you will have to
go to digital servos (down to about 1us deadband), and/or use gearing.


Doubtful, unless you use some precision gearing. If you need
repeatability down to 0.3175 mm you'd also need some optical feedback,
so that you can recalibrate after going to a home position.

Keep in mind that miniature + powerful + precision = MONEY. You can
always find something that will work, but don't expect to eat that
month.

-- Gordon

    --Speaking of indirect means have a look at some of the stuff Carl
Pisaturo's been doing. Here's one really fantastic one:
http://www.carlpisaturo.com/SRP/index.html

--
        "Steamboat Ed" Haas         :  Dedicated to Spinachio,        
        Hacking the Trailing Edge!  :  goddess of spinach..
                          www.nmpproducts.com
                   ---Decks a-wash in a sea of words---

Mechanics is not dead yet? =)
Anyway looks like an impressive alternative way to do things. Esp as it can
rid of any motorweight.

http://en.wikipedia.org/wiki/Image:Industrial_Robotics_in_car_production.jpg

Don't know what you mean by centimeter scale... does this mean you want a
robot to fit in the palm of your hand?

But to answer your question, take a look at the Rhino site, then dig into
more detail on this trainer. It will show you how to hard home, how the
movement is measured and stored then playback, etc. It's the ideal learning
tool. I have one in my shop if you are in So. Cal I'll be happy to let you
play with it long enough to get a sense of what you need to do to design
your own.
Starte here: http://www.rhinorobotics.com/

I'm looking for a robot that is able to access a 160 x 160 mm area. So a
robot that can access a length of 160 mm measured from it's base is enough.
It doesn't have to be fast, but it must keep 0.3 mm precision.

Scanner sled could be used as starting point for xyz table as an alternative
approach to robotic arm.
XYZ board was me initial idea. But I figured that a robotic arm would take
less space.


Unfortunly I'm in Europe .. ;)

<snip>


Check out Como Drills, in the UK. You'll be stuck paying VAT, but the
shipping should be less.

Como sells some of the best yet inexpensive miniature meter-gear motors.
You can certainly pay more, but these are quite good. Use these with a
512 (or so) strike optical encoder and you should be able to achieve
your 0.3mm motor precision. Whether you achieve this precision on the
ground -- rubber tires slip -- is another matter you'll have to work
out.

-- Gordon