DPRG
DPRG List  



[DPRG] [Fwd: RE: Dallas News inquiry]

Subject: [DPRG] [Fwd: RE: Dallas News inquiry]
From: Mr S szinn_the1 at yahoo.com
Date: Thu Jul 7 13:06:55 CDT 2005

Jim, 

I apologize in advance for not having references to
support what I'm writing.

I think that Dale has done an excellent job of
highlighting the major issues that bolster the
statement that current humanoid robotics is mostly
hype. That 20 minute run time is a show stopper, and
would be a show stopper for any technology (car, power
tools, walkman etc.)

The very reason that some will give for using
human-like robots is the reason that they will
initially fail... and fail badly (IMHO). We humans
have perception based expectations built into our
thinking. A dog that meows like a cat is absurdly
shocking. A humanoid robot that cannot perform
functionally like a human will be met with distaste,
perhaps disgust, but certainly with disappointment. If
a walking humanoid robot, that costs tens of thousands
of dollars, falls down the stairs and gets damaged, it
will be widely reported in tech news services as a
failure, and likely prompt a recall. Trying to build a
human form robot that performs as it will be expected
to perform by end users is a train-wreck waiting to
happen. There are many ways for it to fail... think of
a caregiving robot that ran out of power 3 minutes
before it was needed to report/react to a heart
attack? Who would ever insure such activity? What if a
component failure caused the robot to fall over onto
an 83 year old lady, and exacerbate her frailty to a
point of early death?

These are but a few of the 'expected' failure types.
There are hundreds of others. There are safety issues
when even trying to design such a human interacting
machine. What will power it? Will that power plant be
safe? What if the power supply catches fire while it
is in a home with no mobile people? While it is
burning, will it call the fire department? How do you
design a system which is safe to operate around the
infirm and children?

The liklihood that caregiving robots will soon see a
position of current home care givers is very *VERY*
low. I don't mean to say that it will never happen, or
that no one should try, just that there are so many
things that have little to do with robotics that must
be overcome before there will be any such success.
Power is a huge issue. The intelligence factor for
autonomous robots is another huge factor.

On the other hand, human form robots have other uses
that are much easier to see happening. Some examples
would be: tele-operated robots for Chernobyl disaster
work or space exploration, mine exploration, emergency
service and rescue, and the list goes on.

Another very huge factor in commercial production of
robots is <bussword alert> Return On Investment (ROI).
To achieve ROI, the system has to save time, money,
resources, human lives, or all of the above. There is
very little ROI in a robotic photocopier. In fact,
there is little ROI in a photocopier that tells the
user exactly how to fix it when it breaks, certainly
not enough ROI to make it worth selling currently.

Insurance companies will initially ensure that low or
risky ROI, where human interaction is involved, will
not get insurance. That is to say that if there is
risk of human harm, lack of insurance and assured
failures will keep the ROI improbably low for such
robots.

There is high ROI for many types of machines that are
not autonomous, and not humanoid in form, and that
make use of robotics technology. A photocopier that
can inform the maintenance office of impending failure
without intervention of the end user gives high ROI.

As was pointed out, humans are quite used to
interacting with machines that are not human-form at
all. 
The ATM vs. the human teller? 
Online services vs. calling a call center? 
Wireless messaging and Instant messaging vs. using a
phone? 
Robotic mowers and vacuums vs. cleaning staff. 
I suspect that if it hasn't been done yet, someone
will soon use the hospital type robots to wait tables
in a bar or restaraunt soon.

All of that means there is currently very little ROI
for a very expensive to purchase, expensive to operate
humanoid robot.

IMHO, the Aibo is about as close as you get to having
personal computer assistant in autonomous robotic
form. Most of the functions/features that are
desireable in that robot are more like pet-tricks than
functionally useful at this point, but I see them
getting better. Unfortunately, they have chosen to
take the form of a walking dog which is inefficient
where power consumption is concerned. 

There are security robots (wheeled) that sell for tens
of thousands and provide ROI based soley on salary
savings and reliablity in the late night hours. Note
that these robots also use infrastructure for
navigation in most cases, making them useless outside
of that environment. Human form robots have a *VERY*
long way to go before they are going to function as
well as their wheeled cousins. We humans have made it
to the top of the food chain because our bodies and
minds are flexible and adaptable (although that might
be argued in some cases). There is no artificial
intelligence software or systems that would give a
human form robot the same (or even close) level of
adaptiveness and flexability. This means that
currently, what you get in human form robotics is
little more than a life-sized tim-the-toolman that can
only perform a minor fraction of the things that even
Tim can do without ending up in the hospital. I don't
think that many people will readily pay for a 6
million dollar screwdriver! even if it does look like
a man. While the intelligence and abilities of
autonomous robotics remains low in comparison to that
of humans the ROI for humanoid robots will remain low.


If you were to build one, and teach it to mow the lawn
with your $110 gas powered mower, you would have spent
tens if not hundreds of thousands of dollars to do
something that you can get done for a few hundred
dollars with a task specific robot... and still there
are safety issues. Why pay so much for a machine that
you have to keep an eye on so it doesn't harm anyone
or anything? The bang for buck category score for
humanoid robots is not even half of a thumbs up,
though in the coolness category, they get an
enthusiastic 2 thumbs up!


		
____________________________________________________
Sell on Yahoo! Auctions – no fees. Bid on great items.  
http://auctions.yahoo.com/

More information about the DPRG mailing list