Game Development Reference
In-Depth Information
The current posted speed limit
The proximity of the next traffic light
The state of the next traffic light
The proximity of the nearest traffic ahead of us
The speed of the nearest traffic ahead of us
The speed of traffic ahead of the traffic ahead of us
The speed of traffic near us that may help or interfere with our plans (for a lane
change, for example)
Unexpected events to which we may have to react
While all of those components are factors that we would want to build into a
decision on whether to speed up, slow down, apply the brakes, or maintain our cur-
rent velocity, we certainly don't think about them all at the same time. How often
do we make a mental note of our speed, for example? Every few seconds? Most of
us likely check our speedometer even less than we make a mental note of our
speed—perhaps every 10 seconds. How often do we consciously look for the posted
speed limit? More likely in the process of “scanning the world� we happen to notice
the sign and compare it to our mental record of what we thought it was.
While we may make a note of how far away the next traffic light is, we don't
continually monitor its state until it is close enough that we would need to react to
a change. (My father gleefully tells the story of how, when I was three years old,
upon seeing a traffic light at the next corner turn red, I would stop walking… even
if I was in the middle of the block. Obviously my AI was still being debugged at age
three.)
Determining Minimum Granularity
Most of the examples in the above list are addressable by setting different timers for
checking information. One problem with this analogy, however, is that we have a
hard time differentiating between the collection of information and the reprocessing
of a decision. In a computer situation, the information is already there. We aren't
necessarily modeling the sensory system of “checking the speed.� Our driver agent
already “knows� the speed. We aren't modeling the process of looking out the win-
dow at the scenery and occasionally glancing at the car ahead of us or the distant
traffic light. All of that information is immediately accessible. By only processing
the decision calculations occasionally rather than continuously, we are leveraging
the fact that people don't process stimuli that quickly.
 
Search WWH ::




Custom Search