An Idea That Will Rocket Robotics

Did you mean rocket, or torpedo?

--
Joe Legris

Well said!  :-)

I've never understood why some people want robots that look
like humans.  I want one that looks and functions like a ROBOT.

If you want something warm soft & female, wearing japanese
school girl's uniform... move out of your parent's basement and
maybe you'll meet one.



---------------------------------------------------------------
 Men never do evil so completely or cheerfully as when they
   do it from religious conviction.   -- Blaise Pascal          
---------------------------------------------------------------

Like a character out of RUR? They were vaguely humanoid...
or more like....a pick-and-place robot? Could be aesthetically
pleasing I suppose....or like the Mars Rover?

Perhaps you are of the school that believes form follows function.
I know I do- excessive decoration tends to turn me off. An MP-5
looks deadly because it *is* deadly.

However, this person is expressing an idea that he believes has
some sort of potential; most likely commercial success. And
he might not be wrong. There are large numbers of people, websites,
and companies that are heavily drawn to the concept of humanoid
automata and robots. Are they wrong?  Is Rodney Brooks a dolt?
www.ai.mit.edu/projects/humanoid-robotics-group/

I guess your reaction os a manifestation of 'technological racism':
www.codinghorror.com/blog/archives/000675.htm
 
 Just my 2 pennies. So what *is* a Manoi, anyway?

I looked it up on google.

http://www.digitalworldtokyo.com/2006/07/say_ahoy_to_manoi_the_freakyco.php

Doesn't have much room for hacking. Maybe if you lop off the head you
can mount an extra controller board, etc, on top the shoulders.

Regards humanoid or utiliarian robot, etc, it's funny what yanks
people's chains. Lately everyone I know seems to want to build a little
humanoid, or at least a biped, whereas it's obvious that something with
4 or 6 legs could carry much larger loads, and be much more stable over
almost any terrain except for maybe climbing a vertical wall.

My 19-YO nephew saw the following thing, and immediately wanted one ...
and had no idea whatsoever what it takes to build something this
advanced ...

http://www.etek.chalmers.se/~almir/

Likewise, people used to talk about the "next killer app" for PCs. Now
it turns out the big killer app is watching DVDs and playing music. Buy
a 2-Ghz cpu to watch movies on. That's brilliant.

I was shopping for a notebook PC recently, and the only ones they had
in the store had mirror-lcd displays which were great from watching
movies from across the room, but had horrible reflections of both the
overhead lights and also the "back" wall in the store.

Don't people use these things for real work anymore?

It's not about carrying loads; it's about beating the other guy.  I have
to admit, I'm in the same camp.  Robot sumo was a neat idea, but pretty
quickly gets dull; the winners are always some little wedge, and the
match is pretty much over in about 2 seconds.  Robo-One competitions, on
the other hand, are far more varied and interesting.

Also, let's see your 4- or 6-legged robot navigate something like this:

  <http://www.robots-dreams.com/2006/08/roboone_special_2.html>

I'm not saying they couldn't -- I'd just like to see it.  :)

Best,
- Joe

Yeah, too bad. Sounds like someone [namely Greg] made a wedgie, started
beating everyone else, and and the mini-sumo interest waned at FRRC.
That's the problem with contests based on brawn, rather than brain.

I'll look at the Robo-one videos when I get to a high-bw link. That's
quite an advanced contest, and any robots that can do it successfully
are somewhat advanced beyond most of the walkers I've seen in the US.
The japanese are quite far ahead of us on doing such things. [helps,
also, to start with a $2000 robot with digital servos]. EG, the Seattle
Robothon contest is just run down to the end of the track, turn around,
and come back. Speed, not brains, wins. My hexapod can't win it on
time, driving the servos at maximum warp, but if the course were a
little more complicated, it would have a better chance.

BTW, I'll also bet that the bipeds on the page you mentioned will do
the course much better if they can put their hands "down" when
negotiating the obstacles. IOW, act like quadrupeds. Every baby knows
this by 6 months of age. ;-)





.

I was astounded at the knee-jerk reaction to the OP. "If it's not my
idea it
sucks!"

But -yeah- different 'bots for different tasks.


Joe, that was the coolest things I've seen in a while! (Dan, get bb
soon!)
Was it just me, or did Dynamizer seem to shake his head every time he
fell over- it seemed amazingly anthromorphic.

Cheers,
         Tarkin

The bots on that video are really cool.  But, are they R/C bots or are they
totally autonomous?  I can't tell from the little I read on the site but I
assume they are R/C.  Is that right?

--
Curt Welch                                            http://CurtWelch.Com/
curt@kcwc.com                                        http://NewsReader.Com/

Yes, that's right.  Obviously they have to have the motions
pre-programmed (15-20 DOF is a bit much to control manually!), but the
operator selects the motions with some sort of controller.

Best,
- Joe

Dynamizer does shake his head regularly. Sugiura, and his family, are
focused on the creation of robots as avatars. His oldest son, Yuta, is
a sophomore in college and just won a venture capital contest (and
considerable funding) for his proposal to design and build robotic
avatars. Yuta's robot is named RETRO, and won the sprint event at the
same competition.


It depends on the competition. For the ROBO-ONE Eagle they are
operating in R/C mode, though some of the competitors like Maeda
(OmniZero.2) have built in some sensors and limited autonomous
behavior.

Some of the other ROBO-ONE Special events, specifically the sprint and
the ball contests, the robots have to operate completely in autonomous
mode.

For the regular competitions some of the builders have begun
implementing autonomous ability. King Kizer, OmniZero.2, Majingaa, and
several others have the ability to track and attack their opponent. So
far it's pretty primative - good for high scores from the judges during
the qualifying demonstration part of the competitions, but when they go
up against an R/C controlled robot in the ring they usually end up
losing, so most of them aren't using it during the bouts yet.

Lem
Robots Dreams
http://www.robots-dreams.com

I don't think my comment indicated bipeds suck, but I'm still kind of
amazed at how everyone seems to want a "little man" these days :). Joe
can easily get one of those to run the Roboone contest if he wants to
shell out $1500 or $2000 to get something that's already all designed
for him.

http://www.sozbots.com/robo-one/buyone.html

Also, my nephew I mentioned kind of amuses me, because he thinks that a
PC "power user" is someone who has a 2-Ghz machine to play hi-res video
games and watch DVD movies. Power user. Many of the current generation
of expensive toys are really meant for people who are techno-neophytes.
I guess the same people who will spend $1000 to buy a glorified DVD
player will buy a $1500 robot off-the-shelf too.



Shaking its head is the easy part, let me tell you :).

Joe's interest seems to have rapidly gone from the local mini-sumo
contests to one of the hardest hobby robot contests in the world.
There's quite a large learning margin in between. Also, as someone
mentioned, if the contest mentioned actually involves the owners
controlling the robots via R/C link, that's really not very
interesting, especially when you buy an off-the-shelf bot to begin
with. OTOH, "autonomous" biped soccer contests are probably about the
current edge. Now, those are interesting.

Well, sure, but that's out of my budget by a factor of 10-20.  :)  I'll
get there someday, though -- so far I have two servos and a Pololu servo
controller.


Maybe so (though I'm not one of these).


True.  I never said it wasn't hard, nor did I say it was within my
grasp).  I only said it was interesting.

My current efforts are mostly geared at towards making a small modular
robot on wheels, that I hope will someday be able to do something mildly
interesting.  (But that almost certainly won't be mini-sumo; it wouldn't
stand a chance against even the simplest specialized sumo wedgebot.)

Perhaps part of what makes the Robo-One contests more interesting than
sumo is that, by their very nature, the contestant robots HAVE to be
more general than sumo bots.  A successful sumo bot is generally good
for nothing else except sumo.  But a successful Robo-One bot has to be
able to walk, climb stairs, jump, navigate a "rope" (OK, actually pipe),
play soccer, and much more.  Many of the good ones can do even more,
such as rappel down a rope.  A bot that can do all that is a pretty
general robot, and more likely able to entertain party guests (e.g. by
dancing).  That's cool, to me at least.


Well, different people can have different interests I guess, though I
tend to agree that an off-the-shelf bot with stock programming isn't
much of a challenge.  But the vast majority of the Robo-One bots --
especially the ones that win -- are built and programmed from scratch.  
Many of them by kids.  Give these guys some credit; these are miniature
marvels of engineering.

Also, when I ask myself what robots (apart from repetitive-action
factory machines) are actually GOOD for -- i.e., where they are going to
make a significant impact on humanity -- many of the answers I come up
with actually involve teleoperation.  For example, teleoperated robots
could be significant contributors to opening the space frontier, by
doing assembly/maintenance/inspection tasks in orbit and on the Moon
(yes, despite the time lag).  These Robo-One bots are perhaps the most
advanced teleoperated robots yet developed, and I can easily imagine
scaled-up versions of them working in space.  That's interesting to me.

Autonomous contests are very interesting too (I have degrees in
psychology and neuroscience all due to my long interest in AI).  But
realistically speaking, we're not going to have autonomous androids
doing anything much that's useful in near-Earth space (which is my other
main area of interest) any time soon.

Best,
- Joe

You can use an inexpensive mini-sumo style or differential-steered
platform driven by servos [which are easier to control than regular
motors], as a general base to build all sorts of intelligent autonomous
bots.

They use the Kephera platform ["expensive" mini-sumo style] for all
kinds of sophisticated AI work ...

http://www.k-team.com/



Yeah, all of which are already built into the off-the-shelf platforms.



I'll have to check this.



That's an entirely different area.



Apparently NASA has been working on an "android", or at least the arm
portion, to send into space to install replacement gyros in Hubell
telescope, but of course, there will be NOTHING whatsoever autonomous
about it. 100% human control.

Actually, it has lots of room for hacking and expansion. The body
shells are polycarbonate - the same type as they use for the R/C car
bodies - which makes it easy to customize, or to vacuum form your own
shells. There's quite a bit of room under the shells to build in stuff.

When I first got mine, part of the monitor (beta tester) program, I was
a little concerned about how much additional load it could carry, but
after talking to some of the folks here it turns out that the servos
used in the AT01 are the same dimensions, interface, and electrical
characteristics as the higher torque 4000 series line. One of my
friends has swapped out the leg servos already and more than doubled
the available torque.

Lem
Robots Dreams
http://www.robots-dreams.com

Hi Lem. Thanks for the additional information.