Re: [mowbot] Buried Boundary

Sean Scoggins (smscoggi nospam at
Wed, 30 Jul 1997 17:39:46 -0400 (EDT)

On 30-Jul-97 William J. O'Reilly III wrote:
>Here's another idea. How about illuminating the lawn you want mowed with
>some type of light source? Not in the visable range, but something else.
>The "vision" on a mowbot would know where to mow because the grass is "lit
>up," and a sense of "home" can be found by moving to the source of the
>light. One could focus the light to follow straight lines for edges,
>curves, even "don't mow" areas just by putting a different template infront
>of the lens. Another advantage would be that it is a lot easier to change a
>template that to pull up dog wire.
>If the lawn has trees or bushes, put in two or more lights, just designate
>the "home" as, say, a flashing light.

I've been thinking about this for a non-mower project where I need a
group of small robots to stay within a certain region. I have seen
in the past IR floodlights sold for security reasons (ie put an IR lamp
in front of your house and use a CCD camera to watch for intruders; they
see darkness, but you can see them clearly with the CCD.) Of course,
now that I actually want to buy one, I can't find one in the retail market :-)

The advantages to this include unobtrusive and non-permanent boundary
demarkation, as well as cheap and easy IR photodetectors. The disadvantage
is that it is very hard to illuminate square corners for example, also the robot
would cast shadows that would have to be dealt with.

Anybody know if there is a particular wavelength that would be reflected
strongly by green plants? Perhaps some type of spectral "grass-sensor" is
possible? The proper illumination source and an optical bandpass filter
may be a tad expensive, but would make a purely reactive robot (ie see no
grass to the left -> turn right, etc.) work.

just 2cents


Sean Scoggins <smscoggi nospam at> 30-Jul-97 17:39:49