LinuxDevices has an article about
a new surveillance robot. The $30k PatrolBot runs with Embedded
Linux on VersaLogic's
P-III-based VSBC-8 board. PatrolBot is made
by ActivMedia Robotics and
comes in many
configurations. Basically, you get an
autonomous robot that can wander through a building and patrols better
than a security guard. It can listen to speech and respond. It can
find and fetch objects. It can transmit video to a surveillance
monitor. It can avoid objects with its many sonar sensors. You name
it. It even
comes with a docking station to recharge. If it
really works, a security guard for $30k seems seems like a really great
Over engineered, posted 12 Sep 2003 at 06:19 UTC by motters »
Too expensive. I could build a similar robot for a fraction of the
price. In fact something like a cye robot could do the same job.
Not really, posted 12 Sep 2003 at 14:09 UTC by earlwb »
Actually it has to be big enough so that it isn't easily stolen itself.
Plus it has to be robust enough to withstand some vandalism too.
Something you can stomp on to disable isn't going to do it.
FCC and UL approvals are expensive, plus you have to be able to
transmit video and sound for a decent distance too.
Thus it isn't all that unreasonable cost wise. I think it is a little
on the cheap side myself, maybe they are selling it at a loss to get
some market share and exposure.
I also feel it needs to be somewhat menacing in appearance to serve as
a good deterent too.
Packing a 12 guage magazine fed autoloader is a real plus.
Unfortunately, it seems the various states all have laws against using
a mechanical contrivance to trap, injure or kill criminals. We need to
get the laws updated or changed.
I spoke too soon, posted 12 Sep 2003 at 14:12 UTC by earlwb »
The website finally loaded, I spoke too soon, at $30k it's a piece of
junk. $30k should get a 500 pound, titanium or stainless steel
chassis, wheel or tracked robot with a real sense of presence. If you
kick it it'll would break your foot.
ha! you guys think $30k is a lot? i was working on a similar project
when i was at general dynamics robotic systems. try over $100k for
some piece of crap running with less power than my TI calculator. the
thing must have weighed 400lbs, was impossible to program and broke ALL
the time. and that $100k was OUR cost, nevermind what we were selling
the "finished" product for.
something like a cye robot would NOT work. there are lots of other
considerations, like obstacle avoidance, line-of-sight (cye is only 3
inches tall, how much could it see), ability to traverse obstalces,
sensor payload, etc, etc.
check out the mdars-i robot about half way down, and the mdars-e robot
below it. i worked on both of those.
If I see one, I am likely to steal it for the wheelbarrow tires.
But..., posted 12 Sep 2003 at 21:25 UTC by earlwb »
The Midar's robots are most interesting.
But those are military, they have to cost millions or the military
doesn't want them.
That military AGV looked like fun. Looked like a trinocular vision
system on the front for mapping the terrain ahead.
When making a bot to do a security job though you really need to think
practically. For $30K you could just employ a human security guard who
whould be much better at spotting and dealing with intruders. For
automated solutions to work they need to be cheaper than a human doing
the same job and also more reliable and easier to manage. I've worked
in the automation industry for years and reliability is really a
critical factor. If the thing breaks down every six months and
requires complex or expensive repairs then its just not economically
I would like to see people aiming to make robots a consumer comodity,
affordable to the average joe on the street. At present a robot in
every home may sound like a ridiculous proposition, but didn't someone
once say the same thing about PCs?
The Sick LMS 200 laser rangefinder (blue thingie with the opaque plastic
window between the speakers) runs between $5k-$8k depending on the
configuration and who you get it from. Accurate range data is critical
for the localization system to work, so the bot doesn't get lost. The
robotic base runs ~$8k in the configuration shown.
Conceivably all but the laser rangefinder could be hacked together from
COTS components, using CARMEN (GPL) to do
the localization. If only a cheaper laser rangefinder system was