An Evacuation Simulation Based on Boltzmann Distribution - Advanced Research on Computer Education, Simulation and Modeling - page 332

The people will change
their moving directions
once their escaping path
isn't kept back by the
obstacle

The people will change
their moving directions
once their escaping path
isn't kept back by the
obstacle

The people will avoid the
obstacle only when they
are near to the obstacles.

We can hardly see the
congestion near the exit

There is significant
congestion near the exit

There is significant
congestion near the exit

Fig. 3. ( Continued )

It is obvious that everything is treated as a small grid in the Cellular Automata
Model, which limits the true extent of its simulation. It can't represent the ideal avoid-
ing obstacle path. However, it is of the most efficiency among the three models.
Compared with the Cellular Automata Model, the Social Force Model can repre-
sent the process of avoiding the obstacle and the congestion near the exit better. But
the calculating efficiency is very low.
Our model has nearly the same accuracy with the Social Force Model. In the aspect
of the congestion near the exit our model seems does better. What's more, our effi-
ciency is less than half of the Social Force Model (in the same calculating conditions,
the Social Force Model needs to calculate 2089 times, while our model needs 1018
times).
Then we can get the data comparison as Table 1:

Table 1. The data comparison among three models

Congestion near the
exit

The path avoiding
the obstacle

Times need to
calculate

The Model

Social Force
Model

Avoid once the
person finds it

It can reflect 2089 times

Avoids along the
edge of the
obstacle

Cellular
Automata Model

It can hardly
reflect

48 times

Avoid once the
person finds it

Our Model It can reflect 1076 times