DPRG List  

[DPRG] Detecting Water Ahead to avoid the Water trap

Subject: [DPRG] Detecting Water Ahead to avoid the Water trap
From: Randal P. Matthew rmatthew at drivesandcontrols.com
Date: Fri Oct 24 02:45:01 CDT 2003

Just jumping in...

When I was in school we went to the AAAI competition and used a laser to
accomplish range finding and obstacle avoidance.

I was part of the programming team, so I don't remember all the particulars
of the hardware, but it went something like this.

We used a non-visible laser (IR) pointed downward at a motor that rotated a
mirror at 45 degrees (presumably a horizontal beam). 

We then used a filter on a camera to only see the light and mapped processed
captures for the line (moved very fast and I do remember we had to
compensate for blur (intensity). The resulting image was a broken horizontal
line that would help triangulate position to reflected object (knowing the
height of the camera, the right angle etc.

You could do the same thing with a non-horizontal beam to detect steps
etc... You could also incorporate inclinometer and do some fancy trig :)

-----Original Message-----
>From: dprglist-admin at dprg.org [mailto:dprglist-admin at dprg.org] On Behalf Of
W. E. Cole
Sent: Thursday, October 23, 2003 8:34 PM
To: Earl Bollinger; dprglist at dprg.org
Subject: Re: [DPRG] Detecting Water Ahead to avoid the Water trap

It seems to me that the laser idea has merit.

If I'm walking and I see a reflective surface in front of me that's more
than an inch or two wide, I'd probably assume it is a liquid (until I can
get close enogh to see for sure). How about using a laser and testing for
reflection back to the robot. Maybe add a method to sweep the laser
horizontally across an area to determine the size of the reflecting area and
assume anything under "x" size is too small to worry about (either it's a
piece of glass or such a small amount of water that you can ignore it). If
you added a method of sweeping vertically, you could calculate distance and
how long/wide the area was.

As for testing the depth, that would be a problem because unless the robot
was directly over the "depth" to be tested, any signal or method it used
would have to be done at an angle, since the robot would be on the edge of
the water and I presume you would want a way to test the deepest part of the
water, which is unlikely to be at the edge. A simple solution might be to
use a float attached to a contact switch mounted on the robot's body that
triggers once the robot reached a certain depth of water. If the float
doesn't trigger, then just assume it's safe to keep moving forward and if it
does trigger then assume the water is too deep and you (the robot) need to
try a different direction. (I'd probably devise a sensor on some sort of
"feeler" a few inches in front of the robot to test for a sudden drop under


----- Original Message ----- 
>From: "Earl Bollinger" <earlwbollinger at comcast.net>
To: <dprglist at dprg.org>
Sent: Thursday, October 23, 2003 8:34 PM
Subject: RE: [DPRG] Detecting Water Ahead to avoid the Water trap

> Optical, at first looked promising, but there are so many things that
> cause the water to change that it seems to be too complicated at the
> moment. But maybe something will turn up to make me look at it again.
> Anyway, the ideas I got so far are.
> 7.) A laser reflection doesn't look to promising as the different
> terrain would stop the reflection from getting back to the pickup sensor
> or camera. But maybe I'm wrong here.

DPRGlist mailing list
DPRGlist at dprg.org

More information about the DPRG mailing list