This review concentrates on the Mowbot project's requirements (or goals).
The fundamental purpose of the mailing list is to exchange ideas and
experience so we can each build our own Mowbot. This is not going to be a
commercial project. Some people want to learn what they need to forward
their own projects from the list while contributing their experiences;
others want to concentrate on thrashing out a shared design. Discussions
have ranged widely over the components, practicality and philosophy of
robot construction.
It has been stated over and over again that to make progress, we have
to set our requirements. Things that are ruled out can be eliminated
from future discussions. Things that are required or optional can be
specified so that independent design work can proceed. So let us examine
the requirements that have been suggested so far. There are two sorts:
those that determine Mowbot's form and function and those that constrain
the design in some more abstract way. We seem to be doing fairly well
with the abstract requirements, with widespread approval for most.
Here are those which have been discussed and approved by several posts:
* overall completion deadline of Spring 1997
* total parts cost US$500 to US$600 (excluding R&D)
* all parts used must be easily available
* only needs basic hobbyist tools for construction
* construction (once design is done) should take no more than a week
So these should be taken as fixed (priority 1, MUST) requirements.
On the subject of Mowbot's form and function we have rather more
diverging opinions. But on the whole, the divisions relate to matters
of implementation; there are quite a few broad design requirements that
have received widespread agreement:
* autonomous (person doesn't have to direct it)
* independent (no trailing wires)
* cuts grass
* visit all parts of the accessible area in a timely fashion
* quiet and inoffensive to neighbours
* self-maintaining on a day-to-day basis (empties/recharges itself)
* safe (to people, pets, plants and property; no pollution)
* reliable (avoids damage to itself, cutter doesn't jam)
* secure from theft (has alarm/summons help)
* looks cute and has a winning personality. ;-)
These can also be adopted as fixed requirements without fear of argument.
Some other requirements have been suggested and received support but
aren't as clearly uncontentious:
* no need to explicitly define boundaries
* support for defining explicit boundaries
* design should be modular so independent designs can fit in
* support for cutting of multiple areas without re-programming
* operates continuously
* shelters from rain
* copes with mud/debris
* meets specific performance targets (e.g. area/time, max. slope)
* obeys remote commands
Some of these are weaker than those of the first group of requirements
in the sense that a Mowbot which fails to meet one of these will still be
useful. As such, while some could be promoted to requirements (priority
1, MUST), others could just be called suggestions (priority 2, SHOULD)
or dropped completely. More discussion is needed to decide this.
Specific performance targets have received some attention. Some attempts
have been made to specify the total area to be cover, speed of coverage
needed and hilliness and complexity of obstacles likely to be encountered.
More estimates are needed.
There have been many recommendations for implementation of various
features, so many that I am not inclined to describe them all in detail.
Instead, I list the topics and the major threads covered in each:
boundary - how to define and detect the limits of Mowbot's world
paint, tuned coil, IR beams, buried wires
surface detection using ultrasound, IR and colour
charging - how to get charge in to the battery
recharging station, induction
chassis - design of Mowbot's body
buy-in toy or cart, circular differential drive,
4 wheels with 2 steering, centre of gravity, shapes
control - algorithms, behaviours, strategy
random, strategic, mapped, programmed, behaviour driven
cutter - design of grass cutter
spinning blade, spinning cord, oscillating comb,
comb with cutting disc, cylinder blade,
figures for trimmers, calculation of coverage area
location - design of absolute position location system
IR beacons, 40kHz or Polaroid ultrasonic range
micro - choice of microcontroller
HC11, GameBoy, PIC, no need to decide
motors - design constraints and available motor information
speed, size, power requirements, cost, calculations,
cordless screwdrivers, window winders, suppliers
neat - ideas that might not make it, but would be great!
seed, fertilizer, weed-killer, co-operation between
several mowbots, legs, TV remote control, PIR, camera,
sensors for sun, rain, temperature, humidity, pressure
power - design constraints and available battery/solar cell info
solar cell output and cost, gasoline, battery
capacity and cost, charge/run ratio, calculations
sensors - various: range-finding, grass height etc
anti-collision, IR proximity, touch bars, IR grass height,
step/hole detector, ultrasonic distance measuring, tilt
You can retrieve the discussions on all these topics from
http://dragon.swansea.linux.org.uk/mowbot/archive-by-topic/
The consensus seems to be that we should try not to specify these too
tightly, but should instead encourage competing designs that meet the
requirements above.
The one aspect of the design which seems to have wider reaching
implications the rest, and as such warrants early specification, is the
power supply. The quantity of power available severely constrains the
design of some other parts, such as the cutter and the drive motors.
Although many posts have supported solar power, most have expressed
some reservations and been content with the even more widely supported
battery power. A few posts support chemical power. So it seems that
designers of other subsystems can choose work to one of the three power
budgets implied by these three power sources (roughly tenths of amps,
amps or tens of amps).
None of the other topics seem to be in urgent need of reaching a
conclusion, nor does a conclusion seem likely to emerge from any of
them in the near future! There is still plenty of scope for discussion
here too.
Some experiments on cutting technology and ultrasonic terrain sensing
are ongoing and their results will be reported to the list.
Robin.
-- Copyright (C) 1996 R.M.O'Leary <robin nospam at acm.org> All rights reserved. For licence to copy, see http://dragon.swansea.linux.org.uk/mowbot/