[ale] robots

Byron A Jeff byron at cc.gatech.edu
Wed May 22 09:31:21 EDT 2002


> 
> On Tue, 2002-05-21 at 14:32, Byron A Jeff wrote:
> > > 
> > > Byron A Jeff wrote:
> > > 
> > > > point that it'll be much simpler than my current problems. Here are some of
> > > > the issues I have on the table with my current project: a useful robotic
> > > > lawnmower:
> > > 
> > > Gee, why didn't you go with something like a robotic meat cleaver, at 
> > > least then damage would be confined to a single building..
> > 
> > Safety is at the top of my list. I'm planning on using it in a secured
> > enclosed yard that will idle the unit at the slightest detection of
> > movement. The cutting blade is only 9 inches and will be located under the
> > bot. Someone will have to work real hard to get to the dangerous stuff.
> > That's why teenagers are the most dangerous. They think it would be funny
> > to hop a fence, approach a dangerous item and see what it does. My hope is
> > that if it does absolutely nothing when approached, that it'll fail to
> > be interesting.
> > 
> > > 
> > > So, the real question is, how far along are you?  Is it using embedded 
> > > Linux, or some other OS?
> > 
> > Like I said in my post, I'm not even thinking about the brains right now.
> > It'll be microcontroller controlled though, as I have a couple of dozen
> > varieties of PIC controllers and the software expertise to make them sing.
> > 
> > Since it's a single purpose real-time system, there really isn't a need to
> > carry an OS infrastructure.
> > 
> > I'm still doing component testing. The drive motor system works OK in my
> > prototype wooden frame. I'll have to see how well it does once the gas
> > powered generator is mounted. The cutting motor certainly cuts grass using
> > a 9 inch edger blade. However the cuttings gets wrapped up in the motor, so
> > a shield is the next priority.
> > 
> > As for the power system, I have the alternator and the battery, but I'm
> > looking for a cheap working lawnmower engine. Free would be perfect.
> > 
> 
> I'd like to respectfully submit that you're going about this all wrong. 
> You appear to be trying to take the concept of the Snapper/Toro/Lawn Boy
> and substitute human control with machine control.

I know where this conversation is going... Been down this path many times.

>  In so doing, I think
> you're losing sight of the fact that that concept is based around the
> notion of a walking human operator.  Eliminate the human, and there's no
> longer any need for the characteristics of the mower that are only there
> for the sake of the mower operator.  

I used to think that until I actually started testing components.

> 
> If I were going to try to solve the problem of robotic mowing, I
> wouldn't concern myself with anything that was even remotely
> lawnmower-like.  I would fashion a crude walker, six to eight inches in
> length, that works its legs through analog tech and was fronted with a
> kind of low-speed rotary scythe - a turning blade resembling that of a
> food processor that turned against a stationary bypass blade.  I would
> build several of these and make them solar powered.

Well let's take them one at a time:

Walker: The only slightly new idea here. It requires an even more sophisticated 
suspension system than what already is required for a rolling bot. To be stable
it'll need a minimum of 6 legs. Also legs give much benefit on unstable, uneven 
terrain. Though yards are sloped, they generally are graded to be pretty 
smooth. So the perceived benefits are far outweighed by the added complexity.
Walkers are cool technology, but extremely difficult to engineer and control
the  knees, ankles, and feet required to make them usable.

Blade: Blades spin fast for a couple of reasons. First is so that the grass
can be sheared cleanly. Ripped grass tends to get sick. The second reason is
that rotary bladed mowers can handle grass that's several inches higher than
the blade, while reel/scythe style mowers requires the grass to be near the
height of the blade. Also they have serious issues with common items such as
sticks.  So it would not be possible to just throw them out in the yard.

Size: Limits both the surface area that can be cut and the speed at which
it can be cut. And speed still is an element because the grass will look 
uneven if the the mowers cannot keep up with the growth rate.

Solar: A catch 22. First is that you have to increase the size of the solar
panel in order to gather enough energy for the bot to do its job. But 
increasing the size increases the energy requirement. So solar bots spend
a lot of time just sitting and gathering instead of working.

Multiples: Electronic goats in a herd. Now they must be able to communicate
with each other (or a centralized system that know all of their locations)
so they won't get in each other's way. Or as you discussed having real dumb
avoidance/collision behavior.

See current lawnmower technology isn't built in its current fashion simply
to accomodate the human operator. Gas has very high power/weight ratio. 1 
gallon can provide 500 watts of power for several hours. fast spinning blades 
are very effective cutting technology. Wheels work the vast majority of the 
the time. 

Then there's the issue of trying to get the project done in a timely and
somewhat cost effective manner. Engineering from scratch is both cost and
time prohibitive.

That doesn't mean that I won't be look at alternative methods further down
the line. I bought a reel mower and I do want to see how it does under power.
In the end I'd like to see about moving to a battery only platform and having
a battery tanker to shuttle batteries back and forth.


>  I would consider
> myself successful if I can drop these things in the yard in May and
> gather them back up in November, having never had to actually go out and
> mow the lawn in between.  

True. But it's a bar that'll be difficult to reach. One reason that I didn't
limit my power budget is because tasks become very very difficult when power
is limited. This isn't a project that'll be put on the ice sheets of the Artic.
There's plenty of nearby power and spending 2 minutes filling a gas tank is
trivial compared to a significant amount of time sweating out mowing the grass.

The bar you have set is like trying to pole vault over a 25 ft bar. Someday
someone may get there, but it'll take a prohibitively extraordinary effort
to do so.

> 
> Reference http://www.solarbotics.com/.

[ Solarbot behavior trimmed for brevity... ]

> I think if you go the high-speed whirling-blade route, you're going to
> have more unintended consequences than you can stand (many being
> safety-related) and you'll be spending all your time futzing about with
> this thing instead of going about your other business while your yard is
> slowly munched into submission all summer long.

As I stated, the primary safety mechanism is a secured, enclosed space. Fences
and motion/gate sensors. 

My goal is to build with cheap well understood technology. The further afield
one strays, the more difficult and costly it becomes to engineer. I have a
platform that navigates the yard. I have cutting technology that cuts grass.
I have a power plant that generates sufficient power. I have sufficient
sensor and compute power to self actualize the platform. At long last I'm at
the prototype integration phase.

Now is not the time to change horses. Doing so will simply guarantee not
finishing the race.

Thanks for your thoughts.

BAJ

---
This message has been sent through the ALE general discussion list.
See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
sent to listmaster at ale dot org.






More information about the Ale mailing list