Eva Schiffer

Flocks, Herds, and Schools: A Distributed Behavior Model. Craig Reynolds.

Summary:

This is the original paper on independent agent flocking for computer animation. It talks about the precursors to the boids methods, the forces that act upon the agents, and various strategies for creating and using Reynold's flocking model. 

Problem:

How does one make a convincing simulation of natural flock behavior found in nature?

Method:

The basic method used to simulate flocking behavior in this paper is independent agents. Each agent in the flock has the ability to see some of the other agents around it and it choses to move based on where its neighbors are relative to it. All agents view the world from a self centric point of view. They react with some combination of the three basic flock forces: alignment, cohesion, and separation. 

Key Ideas:

Realistic flocking behavior can be created with independent agents who each have a limited view of the world.
Each agent needs only a simple set of rules.
Without spacial data structures the overall algorithm is bounded by the O(n^2) cost of finding which agents are "near" each agent.

Contributions:

This paper was the first to propose a workable flock model that produced realistic results in real time. The use of independent agents is very different than other flocking simulations that came before. 

Questions:

When I first read this paper I had some questions about the mechanics of the flocking forces and collision avoidance. For the most part Reynold's other works answered my questions.