Information Technology Reference
In-Depth Information
ing (used for the investigation of
intra-driver-variability
) and the following
(used for the investigation of
inter-driver-variability
) vehicle. These data are
recorded together with the speed of the experimental vehicle. So far only one
vehicle has been equipped. This vehicle can be used to prove the possibility
of measuring the necessary required variables. The goal is to equip several
vehicles with this sensor system in order to perform measurements with all
these vehicles on a predetermined section at the same time. In a first step,
with the one equipped vehicle 45 test runs with altogether more than 3000 km
on the Autobahn A2 between Braunschweig and Hanover have been carried
out until now.
To compare the microscopic and macroscopic empirical data sets with
the simulation results, the simulation tool must offer the possibility to obtain
simulation data on microscopic and macroscopic level
. At the moment, the
comparison is implemented by the simulation tool AIMSUN. This simulation
offers an established microscopic car-following and lane-changing model as
well as a detailed description of these models. The measurement of macro-
scopic data is by default possible with the simulation tool. The measurement
of microscopic data cannot be accomplished by the standard simulation tool.
Hence, the application interface of AIMSUN is used to extend the function-
ality and to obtain microscopic measurement data from individual vehicles
which can be compared to the empirically observed behavior.
7
Empirical Results
The implementation of the entire data acquisition concept is currently under
development. This section shows first results for the data acquisition and
calibration of the car-following behavior and its validation on macroscopic
level for the headway distribution.
For calibration the Gipps car-following model was applied [4]. A genetic
algorithm and a goodness-of-fit measure based on Theil's U considering speed
and distance was used in order find a proper approximation. In this paper
the results for the parameter b, which describes the most severe braking that
the driver wishes to undertake are presented.
In a first step for the analysis of
intra-driver-variability
of car-following-
behavior a total of 79 car-following trajectories from the equipped vehicle as
a following vehicle were used for calibration. The test driver shows different
values for parameter b for different car-following trajectories, which can be
interpretated as an
intra-driver-variability
. Additionally the parameter values
for speeds of the equipped vehicle lower than 20 m/s are almost all greater
than
m/sec
2
, for speeds greater than 20 m/s the parameter values are all
less than
2
m/sec
2
. These two results show the
intra-driver-variability
and a
speed-dependency of car-following behavior. Amazingly, many previous in-
vestigations regarding car-following behavior have only considered speeds up
to 20 m/s and therefore could not investigate this speed-dependency of driver
behavior [1], [7].
2
