### Summary

This paper address the problem of simulating autonomous characters (aka non-player characters or NPCs in video games) in a life-like and improvisational manner. The author divides the problem in to three main tiers: (1) action selection, (2) steering, and (3) locomotion. The first tier, action selection, deals with high level goals. The second tier, steering, is involved with dividing the higher level goals in to smaller subgoals that are more easily solved. The third tier, locomotion, deals with the actual action used to achieve the subgoals. This paper deals primarily with tier 2, steering.

The author uses a vehicle model as an example and presents common steering behaviors (e.g. seek, flee, path following, wander, etc.). He goes on to show how flocking can be incorporated in to this model as another steering behavior.

These steering behaviors form the basis for more complex, higher level actions. They can be combined sequentially, and sometimes blended together, in order to achieve such goals.

### Problem

To simulate the actions of autonomous characters in a life-like and improvisational manner.

### Methods and Key Ideas

1. The problem of simulating autonomous characters can be broken up in to three tiers: (1) action selection, (2) steering, and (3) locomotion.

2. Steering behaviors form the basis upon which more complex, higher level goals can be accomplished.

3. Steering behaviors can be combined sequentially, or in some cases blended together.