Alt-BEAM Archive
Message #01962
To: Justin jaf60@student.canterbury.ac.nz
From: James Wilson jameswilson@globalserve.net
Date: Thu, 01 Apr 1999 12:47:19 -0500
Subject: [alt-beam] Re: Socer bots/ Hive behaviour
Justin wrote:
>
> > To make bots that work together weather it be as a hive or as a
> > soccer team one thing is certain. Beam alone can not do it. I can
> > think of only one way pure beam bots would work together but it would
> > not be as a team more as a group of individuals with a common goal.
>
> I disagree. That "BEAM alone cannot do it" is very dubious - much of the
> biological functions you described as intelligent are actually the
> result of a collective of BEAM-like systems.
> Study how those biological systems create apparent intelligence out of
> dumb components and you're on the way to a BEAM solution (hence the "B"
> in BEAM :-)
>
> Your soccer example, for example, only consists of one system - an IR
> tag on the ball. Why not also add an IR tag to the players, with all the
> players of a team being the same frequency. Now, wire these three
> sensing systems (fellow team, opposing team, ball) such that the robot
> tries to avoid members of it's own team, but is attracted to the ball,
> and also to opposing team players.
> The sum of the competiting input now results in teamwork:
> Should the ball be in the hands of a fellow player
Deciding if it is a fellow player is higher order decision making
> the team spreads out, but with a tendancy to mark opposing players
Wow that is really hard to do even for humans they must now keep track
of the ball and all of their team mates and all of their oppoinents
>, and a tendancy to stay resonably near the ball,
Very high reasoning involved how should it decide when it is too close
or to far from and opposing player or ball
thus creating a net in case the opposition
> gets the ball.
> Should the ball be in the hands of the opposition, the team converges on
> him, but tends to spread itself around him, limiting his options.
> Should the ball be free, something interesting ensures...
The requirement of the CPU is because many of the reasoning aspects you
suggest just can not be efficiently done hard wired plus we would like
the bots to react to and adapt to the opposition yes. Thus some times a
bot would have to act in a pseudo random type of behaviour i.e. charge
up to the opposition in possession of the ball and then suddenly stop or
back away to gauge the reaction. After several repeated movements of
this type a change in behaviour would have to occur so that the
opposition would be put off 'balance'. If you watch a soccer game there
is a whole lot of reasoning going on at a very high rate of speed. The
bots on the same team would all have to react to the pseudo random
actions of one or two bots to cover the field.
So this brings me back to my initial conjecture that "BEAM bots may be
able to work towards a collective goal but this is not team work" Team
work requires communication and anticipation of team mates and
opponents. To quote the great one "Don't go to where the puck is go to
where it's going to be." Wayne Gretzkey
Wayne is one of the best team players out there because of his ability
to anticipate his team members moves and how his opposition will react
to them. This is not beam. Beam boils down to neurons firing and a
result occurring. Stimulus response this is what higher reasoning is
based on but with no memory and no ability to change BEAM bots can not
act as a team.
James
>
> This system is non-intelligent and needs no CPU or higher intelligence.
> Of course, if you were to actually build it, a lot of fine-tuning and
> extra details would be needed to get it to work, because at the moment
> it's more of a theoretical example than a practical system, but I think
> it demonstrates the point. (eg, the first thing is that it probably
> needs fellow avoidance to have a slightly higher priority that the other
> two).
>
> That it took me all of three seconds to think of a BEAM solution that
> results in teamwork is perhaps also evidence that BEAM is perfectly
> capible in this area, regardless of whether the solution I proposed is
> not a good one or not :-)
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com
Home