Re: [mowbot] test

Byron A Jeff (byron nospam at
Fri, 30 Apr 1999 19:43:53 -0400 (EDT)

> I am as well, although I think yours is the first message I have received
> since subscribing.

It comes and goes in cycles. Usually there is a flurry of activity, then
it goes dormant until the next cycle.

> Anyone out there have a working mowbot that they would like to share
> specs on? I am currently working on one utilizing the following:

I will. Very similar to your's below. Right now my prototype platform is
the top of a wooden 3 legged round table.

> Power: 12v 5A battery

I'm carrying around a 12v 33AHr hog.

> Drivetrain: Two car power window motors with worm gear.

Same. Center mounted so that the circular body can spin on its axis.
I did most of my drivetrain work last spring. Learned quite a few things:

1. An all metal drivetrain is essential. Every wood implement I used failed
at one point or another.

2. That set screws are better than cotter pins for pinning pieces together.
The prototype motor shaft that I drilled through to put in the cotter pin
broke. Next time I'll simply drill a hole through the axel but not the
motor shaft and screw a set screw against shaft.

3. Lock washers are essential. Everything vibrates and it shakes screws loose.

BTW I'm using a couple of Hyundai WW motors....

> Cutting blade: Nylon blade adapter intended for use on string trimmers.

Tried it. Not much success. My current blade is a 7 in circular saw blade.
Seems to do a pretty good job agains tall, tough grass. Learned that with a
smaller drive motor that a different approach is required. First is to
reduce the cutting area. Second is to protect the blade and motor from
spinning the uncut grass on the ground. So I manufactured a housing
that covers the blade. Something like this:


Legend: (P)latform,(M)otor,(S)haft,(B)lade,(H)ousing.

So the blade spins and the housing prevents both cut and uncut grass from
interfering with the motor shaft. Lastly I put a small caster wheel beneath
the whole assembly so that when the platform shifted the housing doesn't
scrape the ground.

> Cutting motor: Not sure yet, but probably a car heater fan motor or
> similar.

I used a radiator fan motor. It has the heft to carry the saw blade, which is
smaller and flat. So the saw blade actually spins faster than the original
fan blade.

> Brains: PIC Microprocessor

Absolutely. Hadn't gotten this far yet but a PIC will be the brains.

> Grass cutting pattern: Random

This is actually a real problem. By necessity the cut path is much smaller than
the path of an ordinary lawnmower. This makes detecting cut areas rather

My gut says that random isn't going to work effectively because there is an
ascetic aspect to a cut lawn, an expected evenness that random cutting just
isn't going cut, pun intended.

Again I haven't gotten this far, but I'm pretty sure I'm going to implement
a real grass height detector and cut uncut grass.

> Obstacle avoidance: ??? Suggestions??? Probably a combination of buried
> wire for perimeter sensing and touch sensors to avoid obstacles in the
> middle of the yard.

This is a subject of wild debate in this forum, and IMHO one of the reasons
why not much practical progress has been made. The goal for quite a few
folks is to drop a unit into unknown territory and let it figure out the
boundaries. A lofty, and to grunts like me, a seemingly unattainable goal.

So I ratcheted down the requirements quite a bit. Simply put I plan on
a completely bounded area and bump sensors. Somthing like the very popular
railroad ties. Anything I wish to protect will be bounded also.

Lately I'm been thinking about mapping technology. I'm planning on experi-
menting with optical beacons, LED's that pulse at different rates that sit on
the boundaries around the yard. I think that with a few of these, that the
mowbot should be able to map its position. If I work it right I might even
ben able to build a self-contained solar-rechargable unit that will only
speak when spoken to.

Another possibility is using the frequency warping coil tags like in the
retail security sensors. The god thing about this is that the Mowbot is the
only active element, the tags are in fact passive, and thus require no power
to operate.

> Seem reasonable?

Emminently. I suggest to anyone that's interested to honestly think a little
less a implement a little more. For example another big topic here is
micropower management. It made my head hurt, so I finally threw out the
power budget and bought the largest 12V sealed battery I could find. I've
been pleased as punch.

Also I'm an advocate of dropping core knowledge here. Many of us a software
or firmware folks. However trying to figure out how to attach an axel to
the driveshaft of a motor is a completly different skillset. I'd like to
talk about the mechanical (mobility and cutting) aspects more and the
planning and sensor aspects less until more folks get a feel of how to
put together a base that moves and cuts.

Just some thoughts. Hope they may be helpful to someone out there...