RWoodward nospam at gnn.com (Ron Woodward) wrote:
> This would only apply if we had no other method of determining location.
> If we use passive beacons or Vision or something else those can easily
> provide the "boundaries" needed to keep us out of the street.
No, it applies even if we do have a position location system.
The point is that if we can sense and respect boundaries *already* in
the environment, then in some circumstances Mowbot will be able to go
to work without any set-up at all. I require this. If circumstances
dictate that we have to define *additional* boundaries some other way,
e.g. by absolute co-ordinates using a location system, or by laying
down buried wires, that gives us extra functionality, but at the cost
of extra set-up defining those boundaries. You require this.
These two requirements don't conflict. On 19 September, I mentioned
both ways of operating, and later made them both requirements. That just
means we need to design in both mechanisms (or delete one or other of
the requirements). Also on 19 September in the section on ``Sensors''
I mentioned various alternatives that haven't been discussed, like
the anti-boundary. If you kept a copy, I suggest you go and review it.
If you didn't, or if you are a newcomer and never saw the original,
you can get that (or any day's) archive of postings from:
http://dragon.swansea.linux.org.uk/mowbot/archive/
mowbot.960919 is 17K.
> We will have to have at least one method to find where we are that does
> not rely on physical boundaries (like between my yard and my neighbors)
Agreed, but it is possible to define a boundary in many ways that don't
depend on knowing absolute location. Buried wire, IR beam and even a
simple fence are all possible additions you could make to the environment
that would define a new boundary, but not tell you where you are.
> If we already have one method of finding where we are that can keep us
> out of trouble I don't see why we need another. We are working to a
> budget sort of.
I'm not sure that we need to, or can, design a cheap enough way of finding
our location accurately enough to implement boundary limits. The best bet
would seem to be an IR beacon triangulation system. But even assuming we
do get that to work, I would still want some much more basic and reliable
detectors as backup. Losing a 400UKP Mowbot just because a positioning
beacon got reflected in a window would be a disproportionate disaster,
so I think we can afford to cover our bets here.
At least in part, the reliance placed on physical boundary and absolute
location sensors will be determined by their relative qualities. So far,
despite a lack of experimental evidence, I suspect that physical boundary
sensors will be accurate and reliable. I have other suspicions about
position location.
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/