Autonomous vehicle ownership and sharing: a demand forecasting approach for the Puget Sound Region and beyond Dr. Chandra Bhat Co-authors: Patricia S. Lavieri, Sebastian Astroza, Felipe Dias, Venu M. Garikapati, and Ram M. Pendyala

MOTIVATION

The Context Autonomous Vehicles: Vehicles that are able to guide themselves from an origin point to a destination point desired by the individual Individual yields full control to artificial intelligence technology Individual decides an activity-travel plan (or tour-specific information) The plan is keyed into the car’s intelligence system The car (or an external entity connected to the car) decides on a routing and circuit to complete the plan

CTR Research Overview Autonomous Technology Connected Technology CARSTOP project Adoption Societal Effects Planning

Did you know? All New 2018 Vehicles Must Have A Back-Up Camera Roughly 200 people are killed each year and another 14,000 are injured in so-called backover accidents, when drivers reverse over another person without noticing him or her. 20 Automakers (99% of the US market) have agreed to make automatic braking standard by 2022 IIHS estimates that automated braking at full penetration would have prevented 700,000 crashes in 2013 (13% of all crashes) “The data show that the Tesla vehicles crash rate dropped by almost 40 percent after Autosteer installation.” – NHTSA Report Tesla expects a 90% drop with Autopilot 2 Tesla is selling insurance in Australia and Hong Kong

Automated Vehicles and Transportation Technology Infrastructure Traveler Behavior

Two Types of Technology Self-Driving Vehicle (e.g., Google) Connected Vehicle AI located within the vehicle AI wirelessly connected to an external communications network “Outward-facing” in that sensors blast outward from the vehicle to collect information without receiving data inward from other sources “Inward-facing” with the vehicle receiving external environment information through wireless connectivity, and operational commands from an external entity AI used to make autonomous decisions on what is best for the individual driver Used in cooperation with other pieces of information to make decisions on what is “best” from a system optimal standpoint AI not shared with other entities beyond the vehicle AI shared across multiple vehicles A more “Capitalistic” set-up A more “Socialistic” set-up

Regular Traffic Conditions PRESENT DAY Regular Traffic Conditions

PRESENT DAY Icy Patch

PRESENT DAY Incident

Lane blocking, traffic slow down PRESENT DAY Lane blocking, traffic slow down

Congestion buildup, late lane changes PRESENT DAY Congestion buildup, late lane changes

Congestion propagation to frontage, ramp backed up PRESENT DAY Congestion propagation to frontage, ramp backed up

Regular Traffic Conditions V2V Regular Traffic Conditions

V2V Icy Patch

Incident: Information propagation V2V Incident: Information propagation

Preemptive lane changing, freeway exit V2V Preemptive lane changing, freeway exit

Re-optimization of signal timing, upstream detours V2I Re-optimization of signal timing, upstream detours INCIDENT AHEAD TAKE DETOUR

Regular Traffic Conditions Autonomous Regular Traffic Conditions

Autonomous Icy Patch

Avoidance of icy patch, no incident Autonomous Avoidance of icy patch, no incident

Traffic slowdown, late lane changing, congestion Autonomous Traffic slowdown, late lane changing, congestion

Autonomous + V2X Icy Patch

Avoidance of icy patch, no incident Autonomous + V2X Avoidance of icy patch, no incident

Information propagation, preemptive lane changing, freeway exit Autonomous + V2V Information propagation, preemptive lane changing, freeway exit

Re-optimization of signal timing, upstream detours Autonomous + V2I Re-optimization of signal timing, upstream detours INCIDENT AHEAD TAKE DETOUR

Shared Autonomous Vehicles (SAV) vs. Private Ownership Chauffeuring household members Shared Autonomus Vehicles (SAV) Acquired by mobility providers (Uber, Lyft, car2go…) Travelers purchasing transportation $/trip $/mile $/minute

Impacts on the Transportation Network and on the Environment PRIVATELY OWNED AV Subsided fares for social inclusion SHARED AV High empty-vehicle-miles traveled Reduced AV owners’ value of travel time Low empty-vehicle-miles traveled Fares control value of travel time Cancel any network operation gain due to AV platooning Increased energy consumption Network operation gain due to AV platooning Reduced energy consumption Increased congestion Low congestion

Possible AV impacts on Land-Use Patterns Live and work farther away Use travel time productively Access more desirable and higher paying job Attend better school/college Visit destinations farther away Access more desirable destinations for various activities Reduced impact of distances and time on activity participation Influence on developers Sprawled cities? Impacts on community/regional planning and urban design

Impacts on Mode Choice Automated vehicles combine the advantages of public transportation with that of traditional private vehicles Catching up on news Texting friends Reading novels Flexibility Comfort Convenience What will happen to public transportation? Also automated vehicles may result in lesser walking and bicycling shares So, will Cost be the main policy tool to influence behavior? Time less of a consideration

Impacts on Mode Choice Driving personal vehicle more convenient and safe Traditional transit captive market segments now able to use auto (e.g., elderly, disabled) Reduced reliance/usage of public transit? However, autonomous vehicles may present an opportunity for public transit and car sharing Lower cost of operation (driverless) and can cut out low volume routes More personalized and reliable service - smaller vehicles providing demand-responsive transit service No parking needed – kiss-and-ride; no vehicles “sitting” around 20-80% of urban land area can be reclaimed Chaining may not discourage transit use

Central Role of Time Use Notion of time is central to activity-based modeling Explicit modeling of activity durations (daily activity time allocation and individual episode duration) Treat time as “continuous” and not as “discrete choice” blocks Activity engagement is the focus of attention Travel patterns are inferred as an outcome of activity participation and time use decisions Continuous treatment of time dimension allows explicit consideration of time constraints on human activities Reconcile activity durations with network travel durations (feedback processes)

In Summary ABM should… SimAGENT does it all and more… Capture the central role of activities, time, and space in a continuum Explicitly recognize constraints and interactions Represent simultaneity in behavioral choice processes Account for heterogeneity in behavioral decision hierarchies Incorporate feedback processes to facilitate integration with land use and network models SimAGENT does it all and more…

RESEARCH QUESTION

Objective To quantify the potential initial demand for AV technology in the Puget Sound Region (PSR) of the State of Washington in the United States. Forecasting process was divided into two phases: Behavioral framework for AV adoption interest based on latent psychological constructs such as environmental consciousness, technology-savviness, and AV- apprehensiveness. We used Census data of the same region to generate a synthetic population for each Census Block Group and then we predicted how the different demands would be geographically distributed across the PSR. This prediction is then translated into a visualization of the results.

About adoption Two forms of adoption were considered: AV ownership and AV sharing. Individuals were grouped into one of four different categories: interested in AV ownership, interested in AV sharing, interested in both options, and not interested in AV adoption.

Behavioral framework

Modeling Individual preferences for ownership and sharing AVs Consumer interest in the adoption and use of AVs Impacts of individual lifestyle preferences, attitudinal factors, and current use of disruptive transportation services. Modeling methodology based on the GHDM approach Data: 2014 and 2015 Puget Sound Regional Travel Study.

Individual preferences for ownership and sharing AVs

data

Data Survey with 1800 individuals in the Puget sound Region AV interest Not interested in AV sharing or AV ownership (68.5%) Interested in AV sharing only (7.6%) Interested in AV ownership only (8.5%) Interested in AV sharing and AV ownership (15.4%) 51% reside in low-density neighborhoods 12% reside in zero-vehicle households and 39% resided in one-vehicle households 14% used a ride-sourcing service at least once in their lifetime 9.2% used a car-sharing service at least once in their lifetime

Lifestyle Variables Green lifestyle Frequency of transit usage Importance of a walkable neighborhood and being close to activities in choice of home location Importance of being close to public transit in choice of home location Importance of being within a 30-minute commute to work in choice of home location

Lifestyle Variables Tech-savviness Smartphone ownership Do not have and do not plan to buy a smartphone (28%) Do not have but plan to buy a smartphone (4%) Have a smartphone (68%) Frequency of usage of smartphone apps for travel information Frequency of usage of GPS

Concerns about Autonomous Cars Type of concern Not concerned Somewhat unconcerned Neutral/doesnot know Somewhat concerned Very concerned Equipment and system safety 6.9% 4.4% 22.2% 26.9% 39.6% System and vehicle security 8.4% 5.0% 26.2% 26.8% 33.7% Capability to react to the environment 6.2% 3.2% 18.9% 22.8% 48.9% Performance in poor weather or other unexpected conditions 6.3% 4.3% 21.5% 26.5% 41.4% Legal liability for drivers or owners 6.4% 4.2% 24.3% 27.4% 37.7%

Behavioral Model RESULTS

Elasticities Variable Not interested AV sharing AV ownership Both Bachelor's degree (base: less than Bachelor’s degree) -2.33% 15.68% 4.94% 1.20% Graduate degree (base: less than Bachelor’s degree) -2.91% 21.77% Age 18 to 24 (base: ≥ 65 years) -14.86% 24.24% -42.86% 118.18% Age 25 to 44 (base: ≥ 65 years) -16.08% 12.12% -10.71% 109.09% Age 45 to 64 (base: ≥ 65 years) -1.22% -- 0.91% Annual income < $25,000 (base: > $75,000) 6.62% -10.67% -20.00% -11.45% Annual income $25-35,000 (base: > $75,000) 3.09% 1.33% -14.12% -6.25% Annual income $35-75,000 (base: > $75,000) 2.94% -12.00% -12.94% Worker (base: non-worker) -4.23% 20.31% 18.06% 6.67% Kids under 5 years old (base: no kids) 2.17% -6.62% 1.41% 2.31% Kids 5-17 years old (base: no kids) 3.04% -7.94% 2.09% 3.30% Experienced carsharing (base: never) 4.29% -40.96% Experienced ridesourcing (base: never) -9.86% 92.31% -17.07% 18.75% High density household census block (base: <3,000 hh/mi2) -5.59% 44.86% -5.96%

Prediction of the Geographic distribution of demand

Population synthesis We need disaggregate household and person socio-demographic data for entire population of Puget Sound Region. We generated a synthetic population by expanding the disaggregate sample data to mirror CENSUS aggregate distributions of household and person variables of interest. Software: POPGEN 1.1 (developed by Arizona State University).

Visualization Three maps in ArcGIS that show number of people interested in AV ownership only, AV sharing only, and both options (ownership and sharing). Maps show the predicted interest in each Census tract.

Interested in AV Ownership only Number of people Average: 323.5 people/census tract Max: 761.2 people/census tract Std. Dev.: 110.8 people/census tract

Interested in AV Sharing Only Number of people Average: 285.4 people/census tract Max: 685.1 people/census tract Std. Dev.: 102.4 people/census tract

Interested in Both Options Number of people Average: 585.9 people/census tract Max: 1,294.0 people/census tract Std. Dev.: 187.2 people/census tract

Conclusions

Conclusions Results show: Individuals with green lifestyle preferences and who are tech-savvy are more likely to adopt car-sharing services, use ride-sourcing services, and embrace autonomous vehicle-sharing in the future. Younger and more educated urban residents are more likely to be early adopters of autonomous vehicle technologies, favoring a sharing-based service model. Individuals who currently eschew vehicle ownership, and have already experienced car-sharing or ride-sourcing services, are especially likely to be early adopters of AV sharing services. GHDM can be used to predict adoption of autonomous vehicle technologies

Conclusions (cont.) The map resulting from our analysis is an important resource for the different stakeholders involved in planning and implementing the future of transportation: Transportation authorities can identify areas of interest for analyzing AV deployment impacts under alternative future scenarios. Mobility providers of ride-sourcing services and carsharing services can identify effective spatial strategies for deploying shared AV systems. Based on behavioral transferability assumptions → Model may be transferred to produce initial forecasts of AV adoption interest for different cities in the country.