Content uploaded by Dimitris Milakis
Author content
All content in this area was uploaded by Dimitris Milakis on Feb 19, 2018
Content may be subject to copyright.
Implications of Vehicle Automation
for Planning
Sivaramakrishnan Srinivasan, Scott Smith and Dimitris Milakis
Abstract The substantial uncertainty associated with the capabilities and deploy-
ment time lines of automated vehicle (AV) technologies makes it difficult to con-
sider AVs in the long range transportation planning process. At the same time,
given current and anticipated resource constraints, the consideration of AV tech-
nology could be critical for developing efficient and sustainable transportation
systems. This paper documents findings from a workshop of modelers, planners,
and researchers on (1) potential uncertainties associates with AV technology and
adoption, (2) its implications for the transportation planning process, and (3) pos-
sible approaches (including immediate steps) that can help address planning under
uncertainty. The workshop was held in the context of the Automated Vehicle
Symposium 2015.
Keywords Long range transportation planning Uncertainties Scenario plan-
ning Pilot projects Education
S. Srinivasan (&)
University of Florida, 365 Weil Hall, Box 116580, Gainesville, USA
e-mail: siva@ce.ufl.edu
S. Smith
Volpe National Transportation Systems Center, U.S. Department of Transportation,
55 Broadway, Cambridge, MA 02142, USA
e-mail: scott.smith@dot.gov
D. Milakis
Department of Transport & Planning, Delft University of Technology,
Stevinweg 1, PO Box 5048, 2628 Delft, CN, The Netherlands
e-mail: d.milakis@tudelft.nl
©Springer International Publishing Switzerland 2016
G. Meyer and S. Beiker (eds.), Road Vehicle Automation 3,
Lecture Notes in Mobility, DOI 10.1007/978-3-319-40503-2_23
287
1 Why Should Planners Care Now? Workshop
Motivation and Organization
Metropolitan Planning Organizations (MPOs) in the Unites States routinely
undertake long-range transportation planning to help shape the land use and
transportation systems of the region that meet the goals of the residents. The time
horizon of these plans is typically 25 years or more into the future. Over the same
period, the automated vehicle technology is expected to be significantly mature and
widely deployed. Yet, little attention has been given to the potentially transfor-
mative impact this technology could have on the long-range transportation planning
procedures and practice. Arguably, one of the key issues leading to the lack of
attention is the substantial uncertainty associated with capabilities and the
deployment time lines of the AV technology. At the same time, given the current
and anticipated resource (finances, land, fuel, etc.) constraints, consideration of the
AV technology could be critical for developing efficient and sustainable trans-
portation systems.
In the context of the discussion presented, the goal of the workshop was to
assemble a group of modelers, planners, and researchers to discuss (1) potential
uncertainties associated with AV technology and adoption, (2) its implications for
the transportation planning process, and (3) possible approaches (including
immediate steps) that can help address planning under uncertainty.
To achieve the above goals, the workshop was broken down into two sessions.
The first session “Technology, policy and user issues”covered key uncertainties for
planners with three brief presentations and audience discussion to identify the most
important driving forces and potential implications of AVs for transportation
demand forecasting and planning. The second part “Scenario development process
and implications for planning”introduced scenario-based planning as an approach
for dealing with uncertainty with a case-study presentation, host-invited response
from regional and state transportation officials, and concluded with an audience
discussion.
Material presented in this paper is drawn from the presentations and audience
discussions at the workshop. It is important to emphasize that the goals of the
workshop, and hence this paper, are to identify key questions from the standpoint of
long-range transportation planning and to suggest initial steps and methods that can
ultimately lead to the development of appropriate planning procedures for an
automated vehicles world. The paper itself does not provide a recipe for AV
demand forecasting and planning.
The rest of this paper is organized as follows. Section 2presents a discussion of
the major uncertainties associated with automation and its implication (key ques-
tions) for planning. Section 3focuses on what planners can do now. Starting with
scenario-based planning as an approach for undertaking a structured discussion
about an uncertain world, the section also outlines the need for pilot studies and
education as immediate needs identified by workshop attendees. Section 4provides
a short summary of the paper and identifies major conclusions from the workshop.
288 S. Srinivasan et al.
2 Major Uncertainties with Automation and Implications
for Planning
2.1 Factors Causing Uncertainty
Three major areas of uncertainty will influence how automation affects the trans-
portation system during the next 25 years. These are technology, policy, and user
attitudes and behavior.
Four technological elements [1] are needed for wide AV deployment: telematics,
ADAS (advanced driver assistance systems), ad hoc communications (DRSC,
LTE-D) and self-driving. Although driver assistance systems exist today, they will
not have transformative impacts on transportation demand, because a qualified
driver needs to still be engaged. One can also envision high automation (SAE level
4; see [2]) operating in certain conditions and on some roads within the next
25 years. However, it is not clear that full automation (SAE level 5), will be widely
available within the next 25 years.
The second factor is policy. Planning and policy are intertwined [3] with the new
urban mobility (shared cars and rides) raising some significant policy issues. At this
point, the United States has been blessed by having comparatively few regulations.
That said, perhaps planners should be more active in addressing policy issues, in
order to make desirable outcomes more likely. A few questions that planners might
consider include the following: How fast does automation happen? What fraction of
the fleet has to automate before you start to see benefits? How does this interact
with funding? What impacts should we expect on vehicle ownership, VMT, and
land use? What does this mean about what we should build in terms of
infrastructure?
The third factor, user attitudes and behavior, will significantly affect how
automation impacts the transportation system [4]. Key questions include (1) will
users embrace automation and (2) will users be willing to share vehicles, riders and
data? Further, planners must always keep in mind the “who”—we traditionally
think of the user as a person or household, but the user might also be a freight
provider or fleet manager. In terms of willingness to adopt automation, important
factors include capital cost, per trip cost, willingness to cede control (driving or
routing), perception of value (safety, convenience, multitasking). These factors will
influence market penetration (percent owned or shared), changes in routing and
changes in travel disutility. In terms of willingness to share vehicles, rides and data,
important factors include relative cost differentials between owning and sharing, use
of vehicles for transport versus storage, convenience of ride-sharing (wait time,
routing), willingness to share in-vehicle space with others (safety, privacy) and data
privacy concerns. Based on several surveys conducted to date, Smith [5] notes that
20–40 % of respondents indicate a willingness to purchase an autonomous vehicle
while 50–70 % of respondents indicate a willingness to ride in one.
Implications of Vehicle Automation for Planning 289
2.2 Key Questions for Travel Demand and Planning
The uncertainties associated with the AV technology and its deployment leads to
several critical questions about its impact on travel demand and planning decisions.
Some of these are discussed next.
Will demand decrease or increase? Several aspects of AV technology and
deployment models might suggest a decrease in vehicular travel demand. For
instance, fewer vehicles might be needed to move travelers, and more so if the
society moves away from ownership of vehicles to subscription to car/trip sharing
services. As fully automated vehicles determine optimal paths for travel, the trip
lengths might decrease. Further, the millennials (who would be the majority of the
traveling population 15–25 years from now) seem to have less preference for
suburban living than previous generations. At the same time, AV technology opens
up new mobility options to children, elderly, and disabled which could potentially
increase overall travel demand. Further, since travel time in automated vehicles can
be productive time, people may become less sensitive to longer trips (assuming that
the cost of these trips are not substantial). Fully automated vehicles may also have
to make several dead-heading (zero occupant) trips adding to the overall travel
demand of the system.
Will capacity increase? AVs promise safety at higher speeds and therefore the
transportation system can be expected to be faster and have fewer incidents.
Further, the headways between vehicles can also be shorter leading to increased
capacity. Signal timing plans can be further optimized (or maybe even eliminated)
recognizing better “reaction times”of vehicles than humans. At the same time,
during the transitionary period, the mixed-vehicle fleet can lead to decreased speeds
and increased headways to allow for safe interaction of human-driven vehicles with
self-driven vehicles. Finally, the performance of AVs in bottleneck situations
(which essentially define capacity) is still unknown.
What level of automation should we be planning for (now)? While discussions
about AVs often (implicitly) focus on a future of fully autonomous vehicles, it is
important to realize that both NHTSA and SAE recognize increasing levels of
automation. Using the SAE levels of automation, perhaps we should be looking at
levels 2/3 (some controls automated, driver’s role is still present) first as opposed to
level 5 (fully autonomous) now?
What does automation mean for infrastructure investment? The investment
decisions we are making today may last 70 years into the future. While automation
may provide opportunities for savings which is important in financially-constrained
times, allocating funds to deal with uncertain AVs may also be a risky proposition.
It is important to consider at least for three major types of infrastructure: roadways,
transit, and the ITS infrastructure. In the context of roadways, it is important to
consider which facilities are most likely to see AVs and correspondingly start
developing those for a future of AVs.
The future perception of transit is of interest. While shared AVs might substitute
for some bus services, high-capacity fixed-guideway heavy rail systems may still be
290 S. Srinivasan et al.
needed on some corridors. Therefore, investments in AVs may be made instead of
investments in transit or to complement investments to improve a transit system. It
is also important to consider the impacts of automation on the profession of
commercial drivers while making decisions.
Finally, the nature of ITS infrastructure needed will also be different for AVs.
Broadly, while current ITS provides information to humans (say a variable message
sign), ITS for AVs should be designed to communicate directly with the vehicles
and also be capable of receiving and processing information transmitted by the
vehicle. Above all, AVs are likely to generate enormous amounts of data and we
should have the infrastructure to be able to store and analyze these.
What about other “Mega-Trends”and Land Use Scenarios? MPOs may be
experiencing their own unique socio-economic-attitudinal trends which need to be
considered as a complement to AV projections. It was also observed that most US
cities have lower densities of development which may not drastically change into
the future. Simulation studies of AVs in low-density areas shows that the occu-
pancy levels of these vehicles might be low, degrading their effectiveness in
reducing travel demand. Thus, AVs must be visualized in the context of the local
land-use trends and policies.
3 What Can Planners Do Now?
This section of the paper discusses some of the steps that planners can take now
towards incorporating AVs in the long-range transportation planning process.
Specifically, scenario planning is introduced as a structured way of dealing with
uncertainty. As the audience noted, while uncertainty in demand forecasting is not
new, the magnitude of uncertainty introduced by AVs is significantly higher and the
consequences of incorrect decision making can also be substantial. Section 3.1
introduces scenario planning and presents a case-study from the Netherlands as an
example of how these may be applied to study planning for AVs. Section 3.2
presents the need for pilot projects and Sect. 3.3 stresses the need for education as
next steps.
3.1 Scenario Planning
3.1.1 Introduction
Scenario analysis can help planners accommodate uncertainty of the future by
identifying different plausible states of the world and possible impacts on the issue
under question [6]. Scenarios offer more flexible and creative outlines of the future
than probabilistic forecasts. In the following section, we present a scenario study
that identified plausible future development paths of automated vehicles in the
Implications of Vehicle Automation for Planning 291
Netherlands and estimated potential implications for traffic, travel behavior and
transport planning on a time horizon up to 2030 and 2050. We first describe the
methodology and then briefly present the scenarios and conclusions about potential
development and impacts of automated vehicles (see Milakis et al. [7] for more
details about this study).
3.1.2 The Dutch Case Study
The scenario study involved experts from various planning, technology, and
research organizations in the Netherlands and was completed in three workshops.
The methodology comprised five sequential steps. The first step involved the
identification of the key factors and driving forces of the development of automated
vehicles in the Netherlands. The impact and uncertainty of those driving forces
were subsequently assessed, followed by the construction of the scenario matrix.
The penetration rates and potential implications of automated vehicles in each
scenario were then estimated. The study was completed with an assessment of the
likelihood and overall impact of each scenario.
Four scenarios were built around permutations of two driving forces of auto-
mated vehicles: technology and policies. These driving forces were assessed as the
most influential and uncertain among the five driving forces identified by the
experts. The scenarios did also incorporate variations of the remaining driving
forces (i.e. customers’attitude, economy and environment). The four scenarios are
the following: (1) AV …in standby, (2) AV…in bloom, (3) AV…in demand, (4) AV…
in doubt.
The first scenario AV…in stand by describes a path where fully automated
vehicles become available in 2030. The technology develops rapidly but govern-
ment policies are not supportive because the Dutch government foresees substantial
risk and negative impact associated with this new transport technology. Thus,
industry takes the lead in the development of automated vehicles. Strongly induced
travel demand, sprawling trends and pressure on conventional public transportation
services follow the introduction of automated vehicles.
The second scenario AV…in bloom describes a path where fully automated
vehicles become available in 2025. Both technology and policies of the Dutch
government support rapid development of this new transport technology. The
positive economic context and other concurrent societal changes of this period (i.e.,
growth of shared economy, environmental awareness movement) drive the demand
for (shared) automated vehicles high. Travel demand management and other reg-
ulatory measures are necessary to curb the rapidly increasing vehicle travel in the
Netherlands.
The third scenario AV…in demand describes a path where fully automated
vehicles become available in 2040. Dutch government policies promote develop-
ment of automated vehicles because it expects major societal benefits (e.g. con-
gestion relief, less accidents) from this technology. Yet, the complexity of urban
environment along with the first fatal accidents hinders technological development.
292 S. Srinivasan et al.
Penetration rate of automated vehicles rises significantly only after 2040. An
important increase in travel demand follows during subsequent years as a result of a
decrease in value of time and an increase in capacity.
The fourth scenario AV…in doubt describes a path where fully automated
vehicles become available only in 2045 but remain a marginal mobility technology
thereafter, mainly available as a premium service. Slow technological development,
lack of supportive policies and skeptical customers form a highly negative context
for automated vehicles.
According to this scenario analysis conditionally and fully automated vehicles
are expected to hit the market between 2018–2028 and 2025–2045 respectively.
The introduction of fully automated vehicles will likely drive demand for this
mobility technology higher. However, the complexity of the urban environment and
unexpected incidents like fatal accidents; bankruptcy or change in strategic prior-
ities of major industry players could significantly influence the development path of
automated vehicles. The introduction of automated vehicle will likely have impli-
cations on mobility in all scenarios. It is expected that the Dutch government will
need to introduce travel demand management measures in three out of the four
scenarios. Finally, experts assessed scenario 2 AV….in bloom and scenario 3 AV …
in demand as the most likely to unfold in the future.
3.1.3 Implications of Scenario Analysis for Planning
Scenarios can assist long-term planning in developing strategies to quickly and
effectively adapt to major and unpredictable challenges of the future. Stead and
Banister [8] suggest that scenario analysis can provide us with insights to avoid
adverse outcomes in the future. Such analysis can also enhance robustness of
transport policy and infrastructure investment decisions. However, development of
scenarios can also be seen as the first step towards a more rigorous examination of
potential impacts of automated vehicles. For example, the results of a scenario
study could provide input to agent-based, activity-based and integrated land use—
transportation models to simulate potential implications of automated vehicles for
travel demand and externalities such as energy consumption and emissions (see e.g.
[9]).
3.2 Pilot Projects
We need a better understanding supported by data of what will happen on the
demand side and supply side of transportation. Therefore, pilot projects on real
behavior will be useful. An audience member noted that it is helpful to think of the
product development principle that states you do not know how someone will use a
product until it is in their hands—not in a test facility, but taken home and used in
daily life. Therefore, several pilot projects each focused on a different aspect of
Implications of Vehicle Automation for Planning 293
behavior would be valuable in collectively providing data to develop policies and
models. A key challenge here is that the technology is not fully available to the
public and researchers for widespread testing. However, we could also gain insights
from data of existing taxi, car sharing and ride sharing services which start mim-
icking aspects of autonomous vehicles and shared-mobility systems. Finally,
another key question is how we can aggregate data at the global level, so that
projects around the world can gain insights from each other. The community should
also identify some universal performance metrics for evaluating pilot projects.
3.3 Education
We need to educate the planning community, including local and state elected
officials. We need to educate the public. However, public education is difficult, and
it is difficult to convince legislators to care about something that is far in the future.
Some agencies are providing education about new technologies within their
agencies, for example Florida Department of Transportation organizes an annual
Automated Vehicles Summit. We need to continue the conversation through
channels such as webinars, the Association of Metropolitan Planning Organizations
(AMPO), Florida meeting, TRB. There are research roadmaps for connected
vehicle and automated vehicles prepared by AASHTO and funded through and
NCHRP pooled study. Another audience member mentioned the American
Association of Motor Vehicle Administrators (AAMVA), as a working group for
states and a good resource on licensing. Widespread education is also important to
ultimately define the roles and responsibilities in addressing AVs at the local, MPO,
state and Federal level.
4 Summary and Conclusions
Despite the considerable uncertainties associated with the capabilities and
deployment time lines of the AV technology, it is important for planners to start
considering AV technology in their long-range-transportation planning process.
This paper first described three major sources of uncertainty as technology, policy
and user behavior. The implications of these uncertainties for transportation plan-
ning were discussed next. The discussion led to the identification of several key
questions that planners have to start thinking about today. The third section of the
paper presented “scenario planning”as a systematic procedure to discuss uncertain
futures. A case study from the Netherlands was presented as an illustration. It is
envisioned that many MPOs will undertake their own scenario planning exercise.
The need for pilot projects to understand behavioral implications of new tech-
nologies was also emphasized. Finally, the need to educate all involved stake-
holders was stressed. All the material presented in this paper is drawn from the
294 S. Srinivasan et al.
presentations and audience discussions at a workshop at the 2015 Automated
Vehicles symposium. The workshop organizers plan to continue to engage stake-
holders via webinars and presentations/sessions at other conferences.
References
1. Misener J (2015) Implications of vehicle automation for planning, presentation at the breakout
session on planning, automated vehicles symposium, Ann Arbor
2. SAE International (2014) Taxonomy and definitions for terms related to on-road motor vehicle
automated driving systems. SAE International, Warrendale, PA. http://www.sae.org/misc/pdfs/
automated_driving.pdf Accessed 12 Oct 2015
3. Mudge R (2015) Implication of vehicle automation for planning—policy links, presentation at
the breakout session on planning, automated vehicles symposium, Ann Arbor
4. Smith S (2015) User issues, automated vehicles symposium, breakout session on planning, Ann
Arbor
5. Smith S (2015) User issues, presentation at the breakout session on planning, automated
vehicles symposium, Ann Arbor
6. Mahmoud M, Liu Y, Hartmann H, Stewart S, Wagener T, Semmens D, Stewart R, Gupta H,
Dominguez D, Dominguez F, Hulse D, Letcher R, Rashleigh B, Smith C, Street R, Ticehurst J,
Twery M, van Delden H, Waldick R, White D, Winter L (2009) A formal framework for
scenario development in support of environmental decision-making. Environ Model Softw 24
(7):798–808
7. Milakis D, Snelder M, van Arem B, van Wee B, Correia G (2016) Scenarios about development
and implications of automated vehicles in the Netherlands. In: Transportation Research board
95th annual meeting. Washington DC, TRB, 2016
8. Stead D, Banister D (2003) Transport policy scenario-building. Transportation Planning and
Technology 26(6):513–536
9. Childress S, Nichols B, Coe S (2015) Using an activity-based model to explore possible impacts
of automated vehicles. In: Transportation research board 94th annual meeting, Washington DC,
TRB, 2015
Implications of Vehicle Automation for Planning 295
