Hi!
>>a-2) Kernel should generate random IDs for units in
>> each simulation. This means that the ID generated
>> for a building in each simulation should be
>> different from the next one.
> Oh what???
> And how do you want to support navigation and path executing for mobile agents,
> if IDs of the path they planed should change in every cycle?
As far as I understand it, Mohsen does not mean to change the IDs every
cycle, but only at the start of each simulation and then leave it this
way for all cycles of this simulation.
So the agent could stay the same, as the IDs stay the same during one
simulation.
The reason for this change is the wish to make preprocessing harder.
> In actual implementation, agents have to
> determine (only) from a history, maintained on their own, from which side they
> entered the blocked Road, to be able plan a path back. Actually it is
> impossible to do so only from information concerning the apropriate Road,
> because PositionExtra is not (even though should me) useful for it.
It is useful - the documentation states, that the block sits in the
middle of the road (middle between head and tail), so the agents are
stopped at the corresponding side of the blockade: If position_extra >
(road.length/2), the agent is nearer to the tail, otherwise it is nearer
to the head. The node the agent came from is the node he is nearest to.
We use this implementation and it works perfectly.
Regards,
Mathias Exner
ResQ-Freiburg Team
This archive was generated by hypermail 2.1.3 : Mon 15 Dec 2003 - 16:59:29 GMT