Hi, my name is Lukas and I am currently in my 3rd year studying Computing Science(Graphics and Games)at university. For my final computing project I have decided to look into the implementation of flocking behaviour and formations. But since the concept of implementing such behaviour is well documented I need to come up with a little twist. Basically something a little different or a new approch to the problem. If anybody has and ideas or maybe some useful links or book recommendations I would be more than happy to hear from you. Thanks Lukas

The flocking systems I''ve read about most often use one set of rules for every member of the flock. Reynolds, for instance, used these 3 qualitative rules to define his now-famous Boids:

"1. Collision Avoidance: avoid collisions with nearby flockmates
2. Velocity Matching: attempt to match velocity with nearby flockmates
3. Flock Centering: attempt to stay close to nearby flockmates"

It might be interesting to see what sort of behaviors emerge from varying- even conflicting- rules. Dewdney''s Party Planner (September, 1987 issue of "Scientific American") included entities with conflicting behavior rules that made for an activ dynamic- you might try something like that on a larger scale.

-Predictor
http://will.dwinnell.com

Here''s an idea... prove Bernoulli''s Equation using a flock of birds/fish!

That is, set up a problem where a flock must enter a pipe/tube of diminishing radius and measure the density and velocity of the flock and the pressure within the flock and against the walls. You''ll need to create flocking rules that attempt to maintain a constant density within the flock. If you want hints as to how to measure the pressure for a flock, holler... but I won''t give the entire answer away, because then you wouldn''t learn why a flock is like a fluid and why flocking works!

Cheers,

Timkin

And then you can prove why the flock in the shower gets hot when someone flushes the flock in the toilet!

Dave Mark - President and Lead Designer
Intrinsic Algorithm - "Reducing the world to mathematical equations!"

How about combining flocking with a simple pathfinding algo? I''m imagining a school of fish swimming through an obstacle course. They have to go through the loop of fire, around the pirate ship, past points A, B, and C, and then return home. I think combining the pathfinding with flocking rules would be rather interesting to see implemented.

I wrote a fish school flocking app last year that had predators chasing the fish. This was nice. After the project was finished a friend and I tried to incorporate a Genetic Algorithm into the creatures in an attempt to evolve fish that were better and better at not getting caught by the predators and also the predators would get better and better at catching their prey. This worked to a limited degree, but we had many problems balancing the environment. ither we would end up with way too many fish or the predators would get too good and eat them all, and then end up starving off. You could try something like that though, it was definitely fun to watch.

ps. you can dl my original flocking app from here at the bottom of the page. This one has no GA though, just the flocking and the predators.

Thanks for all the replies. Since I am really into gaming i think i might implement a litlle game where the player controls the a dog and has to try and control a flock of sheep(just thought of that). The sheep will have to react accordinly to the players actions. Maybe there could me multiple dogs???
Anywayz if you think this idea is good please let me know.

///////
I could maybe also need some pointers on how the behaviour of the flock will be influnced by the player...Will it affect the rules???
///////
Thankx
Lukas

quote:

---

Original post by DGates
This worked to a limited degree, but we had many problems balancing the environment. ither we would end up with way too many fish or the predators would get too good and eat them all, and then end up starving off.

---

Predator-Prey systems are modelled well by the Lotka-Volterra equations. If you look at the phase space trajectory of these coupled DEs, you''ll find that their are only 3 stable points (two are sinks and one is an orbit). They correspond to:
1) Predators eating all the prey and then dying of starvation (zero Predators and zero Prey... the (0,0) solution;
2) Predators dying out and Prey multiplying to infinity... the (inf,0) solution; and,
3) Predators and Prey numbers oscillating on a closed phase space orbit... The initial conditions and parameters of the problem determine the radius of the orbit and hence the amplitude and phase of the oscillation.

For a system such as the one DGates describes, one way to tune it is to record the numbers of Predators and Prey over time, convert this data to a phase space plot and adjust the system parameters accordingly, so as to produce a stable, closed path in phase space. This will give you the requisite parameters for a stable simulation.

As for the sheep and dog idea, sounds like a lot of fun. It could get a bit frustrating for the player. One thought is to throw a few obnoxious sheep into the herd... sheep that try to lead the flock astray and split it up!

Cheers,

Timkin

"For my final computing project I have decided to look into the implementation of flocking behaviour and formations. But since the concept of implementing such behaviour is well documented I need to come up with a little twist. Basically something a little different or a new approch to the problem. If anybody has and ideas or maybe some useful links or book recommendations I would be more than happy to hear from you."

Well, you might like to ask what makes a rebel?

More specifically, what rules do you need to implement in the flock so that rebellious behaviour is possible and maybe, successful, if you define success as attracting a good proportion of the flock away from the mainstream.

Just a thought to explore if you like.




Stevie

Don''t follow me, I''m lost.

I always thought it would be interesting to use the flocking algorithm as a way to model water droplets. You could then add the rule of gravity and crash the flock into a wall and see if you can observe the same crown pattern that water droplets will create.

If you wanted to create a game you could also try controlling a group of jackels to isolate one gazelle from the herd. Just remember- eat the sick and the old first because they are easy targets and tend to be easily replenished by the herd.