>What about using purely mechanical sensors, for instance little flaps,
>which would be deflected by grass. Using microswitch hi/low grass
>could be detected.
One robot lawnmower patent used this idea. Obviously your switches would
have to operate on extremely low gram force. Perhaps the flaps could block
or unblock an IR switch.
>Using potentiometer or varicap the actual height of grass could be measured.
Again the problem would tend to be force.
>BTW, for both mechanical and optical sensors, simple test could be used to
check >if they are still working - back the robot a few feet down already
cut path and >read the sensors. They should indicate low grass.
Yes, that would be a good strategy to employ, perhaps by smapling at a
higher rate you could also pick this up, because normally you wouldn't
detect grass all the time. Each grass detection could increment a register,
no grass detection would clear the register, if the register ever
overflowed that could flag a possible problem and trigger the test behaviour.
Dave Everett
Mowbot Project Website
http://www.idx.com.au/~deverett/mowbot/index.htm