Re: [robocup-rescue-s] KA_HEARSAY

Hi,

I'm some how approving Arash point of view, that if a team reversed engineer
a simulator, it has wasted its time,and we should some how avoid this to
happen ... [to conduct rescuer research in much more valuable fields]

'the state of a simulator the behaviour of which cannot be directly/easily
simulated
by just studying its source-code' [Arash] ==> 'less information is available
to the agents than is provided by the simulators, and that the simulator
behaviour is
either non-deterministic, parameterised or both' [Cammeron].

These are good ideas,But we should be careful in implementing them.
Though It is right that we should prevent agent developers being able to
model simulators by studying source code, the solution addressing this
problem shouldn't create a simulator which can't be modelled even using
data mining, and learning methods.
Therefore, i think hiding simulator state by adding some error to its
corresponding values is a good and an easy implementable solution, but the
error should be less than 2% of actual value [in case of HP +-20 points] .
Also attempts at having more randomized simulator should be carefully done
to avoid the mentioned problem of having a completely random simulator.

Regards
Majid Valipour,
Member of Impossibles team,
Sharif University of Technology.

On 1/30/06, Cameron Skinner <cam@cs.auckland.ac.nz> wrote:
>
> Hi.
>
> I agree with Arash completely. I think the easiest way to secure the
> simulators is to ensure that less information is available to the agents
> than is provided by the simulators, and that the simulator behaviour is
> either non-deterministic, parameterised or both.
>
> In my opinion the simulators that are the least secure are the misc and
> fire simulators. We could fix the misc simulator by rounding the
> PROPERTY_HP and/or PROPERTY_DAMAGE values that get sent to agents, for
> example sending PROPERTY_HP rounded to the nearest 1000 would make it
> much more difficult for an agent to determine the exact state of a
> civilian. The initial state of each civilian is already somewhat
> randomised.
>
> The fire simulator could be fixed with a simple modification that
> randomises the fuel load for each building to some extent. The 3
> fieryness levels already perform the same "rounding" function that I
> suggest for the misc simulator.
>
> With those two changes it might become more expensive for teams to
> attempt to reverse-engineer simulator behaviour than it is to do real
> research on the domain. The Robocup Rescue Simulation League competition
> is there, in my view, to drive interesting research in the fields of
> disaster simulation, multi-agent technologies and artificial
> intelligence, not to teach developers how to read other people's code.
>
> Cheers,
> Cameron.
>
> --
> Cameron Skinner
> Artificial Intelligence Group
> Department of Computer Science
> The University of Auckland
>
> email: cam@cs.auckland.ac.nz
> phone: +64 9 3737599 x82924
> fax: +64 9 3737453
>
> Post:
> Department of Computer Science
> The University of Auckland
> Private Bag 92019
> Auckland
> New Zealand
>
> _______________________________________________
> robocup-rescue-s mailing list
> robocup-rescue-s@mailman.cc.gatech.edu
> https://mailman.cc.gatech.edu/mailman/listinfo/robocup-rescue-s
>

_______________________________________________
robocup-rescue-s mailing list
robocup-rescue-s@mailman.cc.gatech.edu
https://mailman.cc.gatech.edu/mailman/listinfo/robocup-rescue-s
Received on Tue Jan 31 10:36:53 2006