Information Technology Reference
In-Depth Information
Algorithm 1.
Trac mo del
1:
if
attentive
then
2:
v
t
+
3
←
min
v
t
+
a
⊕
(
v
t
)
,v
max
,v
E
max
v
t
+
3
←
min
v
t
+
3
,dist− sd
v
t
+
3
3:
if
random ≤ prob
v
t
+
3
then
4:
max
v
t
+
3
− a
v
t
+
3
,
0
5:
v
t
+1
←
6:
else
7:
v
t
+1
← v
t
+
3
8:
else
9:
v
t
+1
← v
t
10:
position
t
+1
← position
t
+
v
t
+1
drive with continuous velocities. The basic behavior of a simulated car was
described in [5]. Algorithm 1 shows an extended version, which notices brake
lights of the preceding car [8] and a slow start rule [10]. Each simulated car
performs Algorithm 1 in parallel in order to update its velocity. One time step in
the simulation correlates to one second real-time. In line 2, the car will accelerate
from its velocity
v
t
at time
t
with a velocity dependant amount
a
⊕
(
v
t
), if it is
slower than its own maximum velocity
v
max
and the maximum allowed velocity
on the road
v
E
max
. Afterwards, it will brake in line 3, if it is not able to hold the
safety distance to the preceding car under usage of the current velocity
v
t
+
3
.
Line 3 avoids collisions. Line 4 lets the car dally with a probability,
prob
v
t
+
3
,
which is dependent on the current velocity (slow start rule) and on the brake
light of the preceding car. We also assume, that a car on the left lane will dally
less than a car on the right lane, in order to overtake cars on the right lane. Each
car has a standard probability for dallying
⊥
, which is a normally distributed
factor (
μ
=0
.
3,
σ
=0
.
1) as basis for
prob
v
t
+
3
.If
v
t
+
3
= 0 and the car “sees”
brake lights,
will be multiplied with 2. Afterwards, the probability is divided
by the lane used by the simulated car (counting from right to left where the right
⊥
lane is lane 1). Dallying is done by decelerating with an amount of
a
v
t
+
3
.
If a car has
v
t
+
3
= 0, dallying will result in a negative velocity. The resulting
velocity
v
t
+1
thus is limited to 0 as lower bound. The braking without deeper
reason was first described in [13]. It has been widely examined and enables the
simulation model to produce trac jams without external influences.
Lane changing is done w.r.t. an incentive criterion “Do I want to change lane?”
and a security criterion “Is it possible to change without accident?”, according
to [14]. Ways to simulate the occurrence of accidents in microsimulation models
for trac simulation were studied in literature (e.g., [12]). The main idea in these
works is the integration of inattentiveness leading to rear-end collision accidents.
Each car is inattentive with a probability
⊥
and does not change its velocity in
this iteration (see line 9). This may lead to an accident, if the driver in front
