Advertisement

any idea about group movement?

Started by July 12, 2004 09:07 PM
1 comment, last by Lutz 20 years, 4 months ago
hi i'm currently doing a research on group movements. however, i'm confussing in between swarm intelligence, flocking, steering, and move target search, LRTA*, A*. i wish to do a research on the group movements. and implement one of the algorithm there. any idea?? regards, jack
Best Regards,from psjie
One of the easiest methods to move seperate units as a group is to have them follow a limited set of rules:
  1. Don't get too close to any neighbors
  2. Don't get too far from any neighbors
  3. If you are not near any neighbors, actively search for a neighbor.

That's usually the method used in small scale swarm robotics that can sense each other's presence. That will just keep a group together, not give them any direction. In order to achieve direction and order, a couple of methods can be used:
  • Have a leader
  • Have a central base giving commands
  • Borg technology :) Seriously, if they all have the same algorithm and same information, then they should make the same choices.
  • Democracy! Have an ad-hoc network with each unit sharing info and deciding as a group where to go.

The majority-vote idea is cool 'cause if 3/4 of the swarm is closer to peril while 1/4 is safe, the group will decide to high-tail it outta there!

Don't take my word on any of this, though. I only know what works for small robots, not for large combat simulations.
"Donkey, if it were me, you'd be dead."I cna ytpe 300 wrods pre mniute.
Advertisement
I've once seen some movie where they had thousands of rats rushing through the sewerages. To simulate this, they did a fluid mechanics computation of the flow through the channels, then they took the velocity vector field and plugged "rat position points" into it and integrated them along the vector field. So it's like throwing rats into a wild river and look what they do.

This topic is closed to new replies.

Advertisement