The South Korean company Robotis
unveiled RX,
a humanoid robot which can run. The Robotis company is, of course,
the same
company that makes the cool modular Bioloid robot-making kit
that we said would make a
great christmas gift in 2005! Subsidized by the Korean government
to the
tune of around 1 million dollars, RX
is the third robot that can run after Sony's Qrio and Honda's
Asimo. The 2 foot tall RX has 25 joints, can run about 1/2 a mile per
hour, and can even jump. What's really cool about RX, though, is that
it can be reconfigured by its modular parts which made Robotis kits famous.
I'd be interested to know how much this will cost since QRIO was never
available for sale but would have been 'as much as a BMW' I once saw
quoted... I'd also like to see how modular it is, some more pics of it
in bits/made into a dog would be interesting.
It's a good idea to use a PDA as the controller because various O/S's -
Pocket PC / Linux that will probably run on it so it could make the
ultimate hackable / reprogrammable robot product.
I often wonder just how much compliance the joints in robots like these
are designed to have. Saving an expensive mechanical linkage from
damage is one thing, but until truly adaptable compliance that does not
sacrifice accuracy or longevity can be achieved - this problem, along
with that of scalability, will be hugely limiting to the advances that
robots like these can make. Honda, Sony and others have shown that
multi-millions can only go so far to solve this problem - I can only
guess at how much it costs to repair ASIMO when it falls down the
stairs (though I bet they'll never release that mpeg.. ha)
Rigidity is not a desirable characteristic for locomotion - you see it
nowhere in nature, and for good reason. Robots that rely on it for
repeatable accuracy will always be restricted by the very same thing
that makes them excellent at what they do - if they are to 'grow up' -
moving out of showcase project labs and commercial sectors, then this
is one area that has to be addressed - conventional electromechanics
seems unlikely to come up with an answer to this very soon. Pneumatics
are great.. but generating the air supply, and controlling it's flow -
is not. Hydraulics ? - no, same problem - plus the actuators are just
as rigid as before..
Until something truly analogous to muscles is developed, that does not
require a bulky or impractical power source - then robots will not be
doing a decent run, even *after* they can walk really well - a
technological brick wall exists down this path that these kind of
robots are on - and they are slowly, rigidly and expensively..
aproaching it. Perhaps Sony have already realised this ?