[Robotgroup] Drive a Droid proposal... How to detect the "fence"?

Andre Lamothe ceo at nurve.net
Mon Jul 7 19:40:12 PDT 2008


There is going to be a lot of noise, wind, rain even potentially, to get RF 
to work would be too complex, I would stick to basics and use reflective 
tape (like from home depot), and make 3 perimeters around the arena like 1 
foot apart or something, you constantly monitor two ground pointed sensors 
one at the center of the front between the wheels, same in back, then the 
moment you get a reflection with sufficient magnitude, you simply backup 1 
ft, do a rotate 45-90 then move forward 1 ft, if you don't hit anything then 
you release, else you perform another rotate, this will keep it from getting 
out and will "bounce", now if you have a nice compass in there, and you make 
the walls of the enclosure parallel and perpendicular to the w,s,n,e poles 
then you can do a more realistic bounce algorithm and avoidance.

But, reflective tape around the arena, sensors will probably be the most 
realistic and practical (its going to get done) method of doing it, RF, 
lasers, sonar, etc. is asking for trouble. Also, with this technique you can 
make circles or any shape enclosure as long as its not too convex and it 
will work.

Andre'


----- Original Message ----- 
From: "Vern Graner" <vern at txis.com>
To: "The Robot Group Mailing List" <robotgroup at puremagic.com>
Sent: Monday, July 07, 2008 9:01 PM
Subject: [Robotgroup] Drive a Droid proposal... How to detect the "fence"?


Hey all-

I wanted to see if I could get some ideas on how to accomplish something
with the "Drive a Droid" proposal we're submitting for First Night.

We are proposing placing the finished R2D2 droid in an "arena" and
allowing folks to drive the droid and control all its functions (sounds,
lights, movements) via remote-control.

The issue we're trying to figure out now is how do we make it so the
droid can "sense" the edges of its "enclosure" and not go "out of
bounds" when being driven by the "guest driver". The idea would be to
"disconnect" remote control from the driver long enough to reorient the
droid to point away from the arena barrier, then give control back.

The "Drive-a-Droid" First Night proposal (draft!) is here:

http://www.notepad.org/Drive-a-Droid6b.pdf

So you get a feel for concept.  :)

I was thinking about placing a CMU cam on the bottom of the droid near
the "center foot" pointing down and having the droid turn 180 degrees
whenever it saw its "trigger" color. This might allow us to use white or
black gaffer tape to mark off the "area" the droid can roam. Just have
to have a consistent floor color that is high contrast with the tape
color. I figure we could adapt "line follower" or "Sumo bot" type code
for this purpose.

It was also suggested that we use an "invisible fence"-type system with
a low-level RF signal being fed into a ring of wire around the arena.
Have the droid react to the signal.

However, due to the omni-directional nature of the receiving antenna
(without getting into triangulation) we won't be sure exactly what part
of the droid is facing the wire so it would be hard to determine what to
force the droid to do i.e. "turn left" or "back up".

Anyway, I thought I'd throw this out there for some brainstorming. If
push comes to shove, we could create a "2nd set" of controls and an
over-ride switch for a trained operator to use to take over for a
moment, point the droid back to the center and then release control back
to the guest driver.

So, anyone else have ideas on how to make the droid turn away from the
roped area?

Vern

-- 
Vern Graner CNE/CNA/SSE    | "If the network is down, then you're
Senior Systems Engineer    | obviously incompetent so why are we
Texas Information Services | paying you? Of course, if the network
http://www.txis.com        | is up, then we obviously don't need
Austin Office 512 328-8947 | you, so why are we paying you?" ©VLG
_______________________________________________
Robotgroup mailing list
Robotgroup at puremagic.com
http://lists.puremagic.com/cgi-bin/mailman/listinfo/robotgroup 



More information about the Robotgroup mailing list