RE: Implementing vision in Swarm

[Tomas Gudmundsson]

Hi !

We have a similar problem. We want to implement agents in a real valued space 
(no grid, positions are real numbers). This makes the problem of finding 
neighbours not-so-easy-anymore.

As far as I can see (and that is not very far, I am fairly new to Swarm !) 
there might be the following three solutions :

  1. Keep lists of all agents (we do that anyway). Search all lists whenever an 
agent wants to find neighbouring agents. Inefficient, little housekeeping, 
simple to implement.

  2. Make a discrete space on top of the real space. In each cell of this space 
maintain a list of agents contained in that cell. A search for neighbours would 
then be restricted to a few cells, depending on how far we want to look and the 
resolution of the grid. Less inefficient, some more housekeeping, more 
complicated to implement.

  3. If the agents have a physical size, subdivide space into a quad-tree. 
Finding neighbours is then a question of searching the tree. Efficient, a lot 
of housekeeping, probably complicated to implement. This idea comes from Ginger 
Booth address@hidden and I think she implemented it. Hopefully I have 
understood it right ?


Best Regards
Tomas Gudmundsson
Danish Hydraulic Institute