Content uploaded by Nassim Motamedidehkordi
Author content
All content in this area was uploaded by Nassim Motamedidehkordi on Nov 06, 2017
Content may be subject to copyright.
ITS World Congress 2017 Montreal, October 29 – November 2
1
Paper ID SP0792
Impact of Automated Vehicles on Capacity of the German Freeway Network
Martin Hartmann1, Nassim Motamedidehkordi2*, Sabine Krause2, Silja Hoffmann2, Peter Vortisch1,
Fritz Busch2
1. Karlsruhe Institute of Technology, Institute for Transport Studies, Germany
2. Technical University of Munich, Chair of Traffic Engineering and Control,
Germany,*nassim.motamedidehkordi@tum.de
Abstract
This paper shows the results of ex-ante impact assessment of automated vehicles on capacity of freeways
using microscopic traffic flow simulation. The simulation was conducted for different penetration rates of
automated vehicles in Germany´s national vehicle fleet, which were predicted using a newly developed
vehicle cohort stock model. For this aim, the standard segments of German freeway infrastructure including
basic, merge, diverge, and weaving segments were simulated. The resulting capacity increments were
assigned to a country-wide traffic flow model of Germany. In the next step, an economic appraisal was
conducted based on the methodology for the cost-benefit analysis used in the current German Federal
Transport Infrastructure Plan (BVWP). The results reveal that the conservative driving behavior of
automated vehicles, as foreseen by the current legislation, has a negative impact on the capacity of
freeways. On the contrary, automated technologies that allow shorter headways between the vehicles, have
the potential to increase the capacity of the freeway network by 30 % and reduce traffic delays significantly.
However, small market penetration rates of automated vehicles do not lead to discernible capacity benefits
and the potential benefits are likely to be realized at higher penetration into the traffic mix.
KEYWORDS:
Automated vehicles, Capacity, Freeway
1. Introduction
Automated Vehicles (AV) are becoming a reality with a lot of pilot projects demonstrating and testing the
technology abilities as public authorities are opening public roads to test AVs in real traffic. Proponents of
vehicle automation argue that reduced accident rates, increased road capacity and increased social
benefits due to better in-vehicle time utilization may occur. However, in short term these vehicles operate
along the conventional vehicles and their impact on the overall traffic network is unclear. While due to
conservative-assumed driving behavior of AVs in the introductory phase the initial impact on capacity might
be negative, it is broadly expected that major diffusion of AVs within the vehicle fleet will show a
harmonizing, and thus positive, impact on capacity of the transportation infrastructure. From the network
operator perspective, it is important to understand the potential capacity increase in order to address
network planning accordingly. In this paper, the ex-ante impacts of partially-, highly- and connected highly
automated vehicles on the freeway capacity in Germany were investigated.
To be able to estimate the impacts of an unknown amount of AVs on the current infrastructure, the
penetration of AVs within the German vehicle stock was predicted for the next decades and considered
ITS World Congress 2017 Montreal, October 29 – November 2
2
within the capacity analysis. This was done by examining the impacts of AVs on individual types of basic,
merge, diverge, and weaving segments of the freeway network with respect to the predicted penetration
rates. Therefore, a microscopic traffic flow simulation was deployed to model the altered driving behavior
of AVs on the segments typically represented in the German freeway network. Finally, the local impacts
were extrapolated and assigned to the country-wide traffic flow model of Germany (PTV Validate) to draw
economic conclusions.
The paper is structured as follows: First, the literature investigating the impacts of automation on the
freeway capacity is reviewed. Next, the general modelling framework of the capacity analysis and the
assumptions made within the investigation are described in section 3. The simulation scenarios, which are
developed to investigate the impacts on freeway capacity, and the simulation results are illustrated in
sections 4 and 5, respectively. In section 6, the results of the simulation are extrapolated to the entire
freeway network in Germany. Finally, a simplified economic appraisal based on the current method for cost-
benefit analysis used in the German Federal Transport Infrastructure Plan (BVWP) is presented.
2. Literature Review
Much research has been dedicated to impact assessment of AVs on various indicators of traffic flow
efficiency, especially capacity [1,2,3,4,5]; however, these studies focused mainly on longitudinal automation
of vehicles. According to [6,7] automated platooning of vehicles can contribute to substantial influence on
freeway capacity. Vander Werf et al. [8] showed that the longitudinal automation of the driving task can lead
to an increase in lane flow from 2,100 veh/h/ln on a freeway with 100 % conventional vehicles to 2,900
veh/h/ln when the fleet consists of a mix of 20 % conventional vehicles, 20 % Adaptive Cruise Control (ACC)
and 60 % Cooperative Adaptive Cruise Control (CACC) equipped vehicles. Shladover et al. [9] examined
the implications of ACC/CACC technology on freeway capacity through microsimulation with varying
penetration rates of each technology and concluded that CACC can potentially double the capacity of a
freeway lane at a high CACC market penetration rate. However, van Arem et al. [10] illustrated that the
penetration rates lower than 40% do not have an effect on capacity.
Modeling of national vehicle fleet is of research interest due to its suitability for testing various fiscal policies
concerning electromobility, road pricing, pollution zones or most recently autonomous vehicles. In the most
recent works, researchers developed stock-flow cohort models suitable for the simulation of AVs market
penetration. The Norwegian stock-flow cohort model [11] forecasts the Norwegian automobile fleet onto the
2050 horizon in a bottom-up manner to provide a tool for short- and long-term fiscal policy analysis. In
Germany, a vehicle technology diffusion model for Germany and USA was developed to study the impacts
of automated driving on the mobility behavior [12]. The results of the same model served as an input into a
travel demand model to quantify different impacts of AVs introduction [13].
3. Framework and assumptions
3.1. AV characteristics
It is recognized that various definitions are available regarding connected and automated vehicles. In this
research the functions, which correspond to the level 1, 2 and 3 of automation (based on the definition of
VDA [14]), are considered. This study examines three different types of AVs:
In a partially automated vehicle (PAV), only longitudinal control of the vehicle is handed over to the system
and the lateral movement, specifically lane changing, is controlled by the driver. PAV functions such as
ACC are already available in the market. A highly automated vehicle (HAV) is the one in which both
longitudinal and lateral movements are controlled by the system. A connected automated vehicle (CAV)
employs Vehicle-to-Vehicle (V2V) technology by which it can adopt different driving behavior (such as
ITS World Congress 2017 Montreal, October 29 – November 2
3
utilizing shorter headways when following another CAV and more cooperative lane changing) based upon
the surrounding vehicles.
In order to simulate different automation functions, generic models for PAV, HAV and CAV were developed.
The lateral and longitudinal behavior was implemented by changing the parameters of a car-following model
(represented by Wiedemann 99 model) and a lane-change model within the microscopic traffic flow
simulation. An overview of the resulting settings of each AV type is given in Table 1:
Table 1 - Overview of different vehicle types
Vehicle type
Longitudinal
control
Lateral
control
V2V
Headway [s]
Figure
Conventional Vehicle
driver
driver
No
1.1 (with variation)
Partially Automated Vehicle (PAV)
system
driver
No
1.8
Highly Automated Vehicle (HAV)
system
system
No
1.8
Connected Automated Vehicle (CAV)
system
system
Yes
0.9 (when following
another CAV)
Extreme Automated Vehicle (CAV*)
system
system
Yes
0.5
The assumptions for the longitudinal behavior of vehicles are as follows:
AVs obey the recommended maximum speed for passenger cars (PC) which is 130 km/h, maximum
allowed speed for heavy vehicles (HV) which is 80 km/h, and minimum headway recommended by German
Road Traffic Act [15]. As shown in Figure 1, it was assumed that the AVs, unlike conventional vehicles,
have the capability of following their predecessor with a constant headway. Adding the V2V communication
component to HAV facilitates the exchange of information and a better cooperation between the CAVs, by
which lower headways are allowed. Regarding the operation of CAVs, they have the desired time gap of
0.9 s and 1.8 s, if it is following another CAV and a non-CAV vehicle, respectively. In order to investigate
the potential impact of AVs on freeway capacity, an extreme headway setting for CAV (CAV*) was simulated
in which the headways are 0.5 s.
Figure 1 - Headway distribution among the different vehicle types
The assumptions for lateral movement of vehicles are as follows:
First, since the lateral movement of PAVs is monitored and controlled by the driver, the identical model as
for conventional vehicles was used. For HAVs and CAVs, the system takes over the control of lane
changing. For HAVs and CAVs, when compared to conventional vehicles, lower deceleration rates of the
preceding vehicle on the destination lane and bigger safety gaps are allowed during lane changing. Here,
ITS World Congress 2017 Montreal, October 29 – November 2
4
it is assumed that these vehicles follow the German “keep right” rule and return to the right lane as soon as
possible. Besides, both HAVs and CAVs have the cooperative lane change behavior, meaning that in case
the vehicle observes a preceding vehicle on an adjacent lane which intends to change lane, the vehicle
changes its lane in order to facilitate the lane change maneuver execution for the other vehicle.
3.2. Market scenarios and simulated network
In this paper, besides the base scenario (2015, 100 % conventional vehicles), four instances of a trend
scenario (2030, 2040, 2050 and 2050+) and one extreme scenarios 2050+ were covered. The three
instances between 2030 and 2050 correspond to a transition period where the vehicle fleet is represented
by a mixture of conventional and AVs (see Section 3.3. for the specific market shares). Next, the instance
of the trend scenario beyond 2050 corresponds to a saturated AV market. Finally, to explore the maximum
potential benefits of the AVs, the extreme scenario simulates extremely close-headways between the AVs
representing thus as an upper bound within the market scenarios.
Since a microscopic traffic flow simulation of the Germany´s freeway network is impracticable, the network
was split in a number of individual freeway component segments as an input for the simulation. The
definition of these component segments including their capacities is based on the 2015 edition of the
German Highway Capacity Manual (HBS) [16] and includes standard types of basic, merge, diverge, and
weaving segments. To meet the capacities given by the HBS, the component segments shown in Table 1,
modelled and calibrated following guidance from [17], were used for the simulation of the base scenario. In
order to investigate the impact of the AVs on the entire freeway network, the results of individual ´
simulations of freeway component segments are further extrapolated based on the distribution of the
component segments within the freeway network. The location and distribution of the component segments,
determined by using GIS tools, together with the altered capacity values, were transferred to the commercial
network model Validate developed by PTV Group for the extrapolation and economic appraisal described
in Sections 6 and 7, respectively.
Table 2 - Selected HBS freeway segments for the simulation
Label
Component Segment
Label
Component Segment
E 1-2
A 4-2
E 1-3
VR 1-1
E 1-4
S 2
A 1-2
S 3
A 2-3
3.3. Vehicle stock model
In order to investigate the impact of various market shares of AVs on the freeway capacity, a cohort vehicle
stock model was developed to forecast the vehicle automation technology diffusion in Germany in the
upcoming years. As illustrated in Figure 2, the vehicle stock model consists of four component models: (1)
model of PC and (2) HV stock, (3) model of vehicle technology market introduction and penetration and
finally (4) a model of vehicle utilization. Merging these four component models provides a model of a vehicle
ITS World Congress 2017 Montreal, October 29 – November 2
5
technology diffusion in Germany with respect to the trends in car ownership, car utilization and predicted
technology adoption rates.
Figure 2 – National vehicle stock model
The passenger car stock is forecasted based on the statistics collected about the passenger car stock in
Germany in the last 40 years [18]. The key indicators represented by average vehicle age, average vehicle
longevity and forecasted vehicle stock are constrained to match other available forecasts in Germany [19],
besides others under the assumption of an upcoming car peak [20]. Since the share of foreign-registered
HVs on German freeway reaches 40 %, neglecting these vehicles would underestimate the effects of the
automation on the freeway capacity in Germany. The annual road toll statistics published by the Federal
Authority for Goods Transport [21] is therefore a suitable database for the determination of the key
indicators of the HV fleet on German freeways.
Next, a literature analysis provided the estimated years of the introduction of vehicle assistant systems on
the market [22, 23]. In this paper, it is assumed that the German car manufacturers will follow a top-down
approach leading to the introduction of the automated vehicle functions in the premium vehicle segments
first. Further, it is assumed that there will be no regulatory change prescribing a compulsory introduction of
the automation within the manufacturer's vehicle fleet. Through an introduction of the vehicle technology
within the newly registered vehicles in the stock model, the technology diffusion is then a result of the aging
process of the vehicle stock with respect to the annual scrapping and survival rates.
Finally, it is assumed, that the AVs will be overrepresented on the freeways as the frequent car users tend
to spend more money on the assistant systems and generally adopt technology faster. To reflect on this, a
passenger car utilization model was fused to the vehicle stock model. The data for the car utilization model
was collected by the household travel survey German Mobility Panel (MOP) [24] and modelled to obtain
among others the average annual and daily vehicle mileage on the freeway [25]. As a result, Table 3 gives
the aggregated penetration rates of AVs (PC and HV) within the vehicle fleet in Germany.
ITS World Congress 2017 Montreal, October 29 – November 2
6
Table 3 - Penetration rates of each vehicle type in market scenarios
Scenario
Year
Conventional
PAV
HAV
CAV
CAV*
Base (lower bound)
2015
100 %
-
-
-
-
2030
74 %
11 %
9 %
6 %
-
2040
32 %
28 %
22 %
18 %
-
2050
12 %
35 %
27 %
26 %
-
Saturated AV market
2050+
-
-
-
100 %
-
Extremely close-headways (upper bound)
2050+
-
-
-
-
100 %
4. Simulation Experiment
In order to quantify the potential impacts of AVs on freeway capacity, a wide range of simulation
scenarios, with help of microscopic traffic flow simulation Vissim 7.00, was covered:
● Different freeway segments,
● Different types of automated vehicles,
● Various penetration rates of AVs for 2030, 2040, 2050 and 2050+,
● Different loading ratios: the ratio between merging, diverging and weaving volumes and volume
on the freeway mainline.
The general simulation approach within this study is illustrated in Figure 3.
Figure 3 - Simulation approach within the study
In order to account for the stochastic nature of traffic, in each simulation scenario multiple runs with different
“random seeds” were simulated. For each scenario, the number of necessary simulation runs was
determined based on the statistical method provided by FGSV (The Committees of the Road and
Transportation Research Association) [25].
ITS World Congress 2017 Montreal, October 29 – November 2
7
The capacity of each road segment was calculated with the help of fundamental diagrams and the method
used in [17]. In this study, in order to estimate the capacity, the average breakdown volume in 5-minute
intervals was determined. A breakdown was defined as a speed drop below a threshold speed. The
threshold speed was individually defined for each analyzed segment based on the gap between the
simulated data points representing free flow and congested traffic conditions in the flow speed diagram and
ranges between 60 and 80 km/h. To determine the capacity of the freeway segment by identifying traffic
breakdowns in the simulation, it was necessary to increase the demand beyond the capacity of the
segment. For this aim, the capacity values from HBS were considered as a reference. In case that due to
the new behavior of AVs, no breakdown occurred in the simulation, another method was employed for
capacity estimation. In this method, if the traffic volume did not increase for three consecutive time intervals,
despite the increasing demand, the data point close to the 90th percentile of the traffic demand value was
considered as capacity.
5. Simulation Results
This section describes the simulation results, with the focus on the capacity difference of each simulated
freeway segment in comparison to the base scenario. Figure 4 represents examples of the capacity
increments within each scenario relative to the base scenario, for different road segments. The figure
depicts, the capacity deteriorates due to a gradual increase in penetration rates of PAVs and HAVs, whose
behavior is conservative compared to conventional vehicles, regardless of the road geometry. With an
increase in penetration rates of AVs, let it be partially, highly or connected highly automated vehicle, the
negative impact becomes more apparent. Despite 26 % of vehicles being CAVs in 2050, their positive
impact on the capacity is negligible which confirms the fact that the benefits of CAVs are not significant in
low penetration rates. The capacity benefit can be realized to a greater extent when automated technology
is combined with connected technologies and when extremely short headways are allowed as in scenario
2050+ upper bound. The automation of vehicles has the potential to increase the capacity by 45 % in basic
road segments. In weaving segments, where a merge and diverge are in close proximity, intensive lane
changing has a major impact on capacity and cooperation between vehicles is of high importance. That
explains the fact that the negative impact of conservative PAVs and HAVs on capacity in weaving segments
is higher when compared to basic road segments as well as merge and diverge segments. Likewise, the
CAV* keeping the extremely short headways does not provide sufficient safety gaps for lane changes,
therefore the capacity improvement for weaving segments in 2050+ and 2050+ upper bound scenarios is
not as much as in abovementioned road segments.
Figure 4 – Simulation result representing the percentage change in capacity of each freeway
segment relative to the base scenario
ITS World Congress 2017 Montreal, October 29 – November 2
8
It is noteworthy to mention that in diverge segments, due to large headways of PAVs and HAVs, the vehicles
leaving the freeway have the possibility to reach the off-ramp without causing perturbation in mainstream
traffic, therefore in diverge segments traffic was harmonized and fewer traffic breakdowns were observed
in the simulation.
6. Extrapolation to a country-wide traffic flow model
The objective of this section is to give a rationale for extrapolating the simulation results of individual
component segments to a country-wide traffic flow model. The traffic flow model is represented by the
commercial product PTV Validate, which covers the major roads in Germany and splits the country in over
10,000 zones using over 3.5 Mio links. The calibrated PC and HV demand matrices are used to calculate
the traffic flows and are compared against the defined capacities to obtain volume/capacity ratios. Changing
the default capacity values of component segments to the simulation-based results alters necessarily the
network performance indicators. This section presents the results of the extrapolation and sums up the
impacts of the AVs on the mean capacity, total delay, travel time and volume/capacity ratio.
6.1. Capacity
The mean percentage change of nominal capacity is displayed in Figure 5. The figure further shows the
market shares of respective automation levels as defined in Section 3.1 and the resulting impacts on the
capacity comparing to the base scenario (red line). The percentage change of nominal capacity represents
an aggregate value across all individual component segments on German freeway network. The results of
the extrapolation show, that an increasing share of partially- and highly automated vehicles that drive rather
conservatively and “by the book” will result in a capacity decrease up to 7 %. Only with a significant share
of connected and automated vehicles that maximizes fully the potential of vehicle communication resulting
into close-headways and cooperative maneuvering, there will be a significant increase of road capacity
leading to about 30 % capacity increase. These results seem to be rather conservative, however, are
perceived as plausible and in accord with the adopted assumptions.
Figure 5 - Impacts of AVs on the freeway capacity in Germany
0%
20%
40%
60%
80%
100%
120%
140%
2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070
market share / capacity change
Conventional HAV CAV Capacity
ITS World Congress 2017 Montreal, October 29 – November 2
9
6.2. Other performance indicators
Adjustments of the default capacity values in the traffic flow model lead inevitably to changes in
performance of the transportation network. Figure 6 depicts results aggregated over all freeway links within
the Validate model. The freeway performance trends clearly mirror the trend in capacity changes: initial
negative impacts of the automation are observed. First by around 2050 significant positive impacts on the
network performance will be reached.
Figure 6 - Impacts of AVs on total delay and travel time within the freeway network
In Figure 7 base scenario and upper bound scenarios are visualized on the example of volume/capacity
ratio: the majority of bottlenecks within the network were removed in the upper bound scenario and the
overall load of the network was decreased. Needless to say, neither traffic demand growth nor rebound
effects such as rerouting or induced traffic demand are incorporated within the methodology of this study.
Figure 7 - Volume/capacity ratios: Base (left) vs. Extreme scenario (right)
5%
21%
32%
-53%
-63%
-80%
-60%
-40%
-20%
0%
20%
40%
2030 2040 2050 2050+ 2050+
(upper
bound)
Total delay vs. Base
1%
2% 4%
-6% -7%
-8%
-6%
-4%
-2%
0%
2%
4%
6%
2030 2040 2050 2050+ 2050+
(upper
bound)
Travel time vs. Base
ITS World Congress 2017 Montreal, October 29 – November 2
10
7. Economic appraisal
This section presents an economic appraisal of the impacts of the AVs on the capacity and thus on the
travel time in the German freeway network. The appraisal is based on the methodology developed for the
current Federal Transport Infrastructure Plan 2030 [26] that defines the prioritizing strategy in the transport
infrastructure maintenance, expansion and new construction process. In this paper, we base our appraisal
on the cost-benefit analysis (CBA) as the core component of the BVWP appraisal process. The CBA
captures 12 various benefit components, from which three are selected in this appraisal as they are
dependent on travel time changes: Travel time changes in private transport (NRZ), Transport time changes
in heavy goods transport (NTZ) and Changes in operating costs (NB). The travel time reliability component,
as well as the traffic safety component of the CBA, were omitted in this paper. Due to the extent of the
methodology, we refer to the BVWP 2030 and present only the results of the CBA components.
Table 4 - Aggregate annual economic appraisal based on the German BVWP
Scenario
Year
VHT
[M veh∙h/yr]
VHT change
[M veh∙h/yr]
Σ NRZ
[M €/yr]
Σ NTZ
[M €/yr]
Σ NB
[M €/yr]
Σ
[M €/yr]
PC
HV
Base (lower bound)
2015
1,727.7
-
-
-
-
-
-
Trend
2030
1,736.4
0.011
0.001
-162.1
-2.2
-27.1
-191.4
2040
1,774.9
0.050
0.003
-694.7
-9.8
-120.5
-825.0
2050
1,800.7
0.076
0.005
-1,052.4
-15.7
-189.3
-1,257.4
2050+
1,596.8
-0.127
-0.005
1,770.1
15.4
231.0
2,016.5
Extreme (upper bound)
2050+
1,573.3
-0.151
-0.005
2,096.2
16.9
263.3
2,376.4
Table 4 gives the estimated annual economic benefits of AVs based on their impact on the capacity of the
German freeway network. Importantly, this appraisal is closely tied to the assumptions made and a rather
conservative setting of both PAVs and HAVs, basically responding to the driving behavior allowed by the
current legislation. The economic appraisal follows the trend of capacity increments and yields negative
economic benefits during the introductory phase of the automation (2015–2050). However, large economic
benefits can be achieved once high penetration rates of closely spaced automated, and particularly
connected automated vehicles, contribute significantly to the traffic flow harmonization.
8. Conclusions
This paper presents a study about the impact assessment of AVs on the capacity of German freeway
network with the help of microscopic traffic flow simulation. Simulation results clearly convey that the impact
of vehicle automation on capacity depends strongly on the driving behavior of AVs, especially the
longitudinal behavior, as well as their penetration rate within the vehicle fleet. The change in freeway
capacity is not profound in low penetration rates and the potential benefits are likely to start kicking in at
higher penetration rates. Moreover, the capacity deteriorates in case the conservative driving behavior of
AVs is introduced. Further analysis showed that exploiting the potential benefits of AVs (up to 30 % increase
in capacity for the entire freeway network) and better use of road infrastructure will be achieved once the
headway of these vehicles falls below the value in current regulations.
Even though many car manufacturers state their readiness to mass-produce AVs in the next years, large
market penetration of AVs is decades away and the potential impact of the automation cannot be fully
exploited in the near future and with the existing regulations. By then, there might be significant changes in
the current legislature, economic situation (sharing economy) as well as mobility behavior. As a result, our
approach makes a number of assumptions relating to the driving behavior of AVs, structure of the vehicle
ITS World Congress 2017 Montreal, October 29 – November 2
11
stock and market penetration rates as well as the adoption of the new technology. Therefore, it is necessary
to understand the results of this study are in a very close context to the assumptions made. Since the
abovementioned aspects of AVs are associated with significant uncertainty, additional research is
necessary to better anticipate the impact of AVs.
Acknowledgement
This research was financially supported by the Research Association of Automotive Technology (FAT),
working group 7 „Optimization of Road Traffic Systems“. Thanks go to PTV Group AG for providing the
Vissim simulation tool license.
References
1. Kesting, A., M. Treiber, D. Helbing (2010). Enhanced intelligent driver model to access the impact of
driving strategies on traffic capacity, Philosophical transactions. Series A, Mathematical, physical,
and engineering sciences, vol.368, (1928), pp.4585–4605. Accessed January 31, 2016.
2. Jerath, K., S. N. Brennan (2012). Analytical Prediction of Self-Organized Traffic Jams as a Function
of Increasing ACC Penetration, IEEE Transactions on Intelligent Transportation Systems, vol.13, (4),
pp.1782–1791.
3. Bierstedt, J., A. Gooze, C. Gray, J. Peterman, L. Raykin, J. Walters (2014). Effects of Next-
generation Vehicles on Travel Demand and Capacity: FT Think.
4. Atkins (2016). Research on the Impacts of Connected and Autonomous Vehicles (CAVs) on Traffic
Flow, Stage 2: Traffic Modeling and Analysis,Technical Report: Department of Transport.
5. Shelton, J., S. Samant, J. Wagner, G. Goodin, E. Seymour, T. Lomax (2016). Revolutionizing Our
Roadways, Modeling the Traffic Impacts from Automated and Connected Vehicles in a Complex,
Congested Urban Setting: Transportation Policy Research Center.
6. Karaaslan, U., P. Varaiya, J. Walrand (1990). Two proposals to improve freeway traffic flow,
California Partners for Advanced Tranit and Highways (PATH): UC Berkley: California Partners for
Advanced Transportation Technology.
7. Ioannou, P. A. (1997). Automated Highway Systems, Boston, MA: Springer US.
8. Vander Werf, J., S. Shladover, M. Miller, N. Kourjanskaia (2002). Effects of Adaptive Cruise Control
Systems on Highway Traffic Flow Capacity, Transportation Research Record: Journal of the
Transportation Research Board, vol.1800, pp.78–84.
9. Shladover, S., D. Su, X. Y. Lu (2012). Impacts of Cooperative Adaptive Cruise Control on Freeway
Traffic Flow, Transportation Research Record: Journal of the Transportation Research Board,
vol.No. 2324.
10. van Arem, B., C. J. G. van Driel, R. Visser (2006). The Impact of Cooperative Adaptive Cruise
Control on Traffic-Flow Characteristics, IEEE Transaction on Intelligent Transportation Systems,
vol.7, (4).
11. Fridstrøm, L., V. Østli, K. W. Johansen (2016). A stock-flow cohort model of the national car fleet,
European Transport Research Review, vol.8, (3).
12. Kröger, L., T. Kuhnimhof, S. Trommer, eds. (2016). Modeling the impact of automated driving –
private autonomous vehicle scenarios for Germany and the US., Presented at the European
Transport Conference, Barcelona: Spain.
13. Trommer, S., V. Kolarova, E. Fraedrich, et al. (2016). Autonomous Driving, The Impact of Vehicle
Automation on Mobility Behaviour, Munich, Germany: ifmo.
14. Verband der Automobilindustrie (VDA) - Association of the Automotive Industry (2015). Automation,
From Driver Assistance Systems to Automated Driving.
ITS World Congress 2017 Montreal, October 29 – November 2
12
15. Strassenverkehrs-Ordnung (StVO) - Road Traffic Order. BGBl. I S. 1573: Federal Ministry of Justice
,Germany, https://www.gesetze-im-internet.de/stvo_2013/. Accessed January 25, 2017.
16. Forschungsgesellschaft für Straßen- und Verkehrswesen (FGSV) - The committees of the Road and
Transportation Research Association (2015b). Handbuch für die Bemessung von
Straßenverkehrsanlagen (HBS), Edition 2015, (German Highway Capacity Manual), Cologne.
Germany: Forschungsgesellschaft für Straßen- und Verkehrswesen (FGSV).
17. Geistefeldt, J., S. Giuliani, P. Vortisch, et al. (2014). Assessment of Level of Service on Freeways by
Microscopic Traffic Simulation, Transportation Research Record: Journal of the Transportation
Research Board, vol.2461, pp.41–49.
18. Kraftfahrt-Bundesamt (KBA)-Federal Motor Transport Authority (2015). Fahrzeugzulassungen (FZ):
Bestand an Kraftfahrzeugen nach Fahrzeugalter (The annual count of the stock of vehicles
differentiated by the vehicle age), Flensburg.
19. Intraplan Consult GmbH (2014). Verkehrsverflechtungsprognose 2030, Zusammenfassung der
Ergebnisse, Berlin: Bundesministerium für Verkehr und digitale Infrastruktur.
20. Prognos AG (2014). Shell Passenger Car Scenarios for Germany to 2040, Hamburg, Germany.
21. Bundesamt für Güterverkehr (BAG) - Federal Authority for Goods Transport (2015). Mautstatistik:
Jahrestabellen 2015 (Annual Road toll statistics for 2015), Cologne. Germany.
22. Maurer, M., J. C. Gerdes, B. Lenz, H. Winner, eds. (2016). Autonomous Driving, Technical, Legal
and Social Aspects, Berlin, Heidelberg: Springer Berlin Heidelberg.
23. European Road Transport Research Advisory Council (ERTRAC) (2015). Automated Driving
Roadmap, Brussels: ERTRAC.
24. Weiss, C., B. Chlond, T. Hilgert, P. Vortisch (2016). Deutsches Mobilitätspanel (MOP) –
Wissenschaftliche Begleitung und Auswertungen, Karlsruhe, Germany: Alltagsmobilität und
Fahrleistung.
25. Chlond, B., C. Weiss, M. Heilig, P. Vortisch (2014). Hybrid Modeling Approach of Car Uses in
Germany on Basis of Empirical Data with Different Granularities, Transportation Research Record:
Journal of the Transportation Research Board, vol.2412, pp.67–74.
26. PTV Group, TCI Röhling, Hans-Ulrich Mann (2016). Methodological Manual for German Transport
Infrastructure Plan 2030, FE-Projekt-Nr.: 97.358/2015, Karlsruhe, Berlin, Waldkirch, Munich.