Reading Ken's article about his robot and finally reading the Solar mower
patent I have realised that I'm guilty of what I often accuse other
roboticists of, namely complicating their designs.
I've always felt that designing a robot to completion on paper was a recipe
for dissapointment. A better approach is to get a basic platform going then
gradually solve smaller problems until the project is successful.
Grassy started this way. I even used DPDT relays to control the direction
of the motors, a FET to apply motor power and a FET to switch the relay
direction. Normally I would use a PWM motor driver chip like the LM18200
for a project like this, but I wanted to save some dollars (the LM18200's
are $40 a piece) and get the base up and running quickly so I could start
on the problem solving. Somewhere along the way I fell into the engineering
abyss. I was halfway through designing a processor board that provided 16
inputs, 16 outputs, 8 analog inputs, 8 sonar distance measurement inputs,
and RS232 interface and a windows program that allowed me to write
behaviours in a graphical interface and download them to the processor
board. What the hell was I thinking!!
The Solar mower is elegant, solving most of the problems in an efficient
manner. Ken takes that approach too! So I've decided to get back to basics.
It seems the first task to solve is boundary marking, and the simplest
solution seems to be a buried wire. The initial work required to setup the
wire is considerable, but it requires no further maintenance and can be
monitored for failure quite easily. Also with the addition of several
pickups on the robot it would be possible for the robot to straddle the
wire in order to cut the boundary, travel to other areas, return to the
charging base etc.
I'm planning on using a PA line transformer to pump a 17khz signal (or
thereabouts) around a wire. This should give me 100v at about 2ma, a safe
level. I'm very intersted in hearing everyones thoughts on the pickup
system. Bought some inductive pickup coils from Tandy (Radio Shack to you
Yanks) however they work very poorly. I only want to detect the wire at a
distance of about 8 inches. If someone has already built a system or is
working on one let's hear about it.
Dave Everett.