1.
ASSIMILATING SELF‐DRIVING CARS
INTO SOCIETY:
Travel Impacts & Transport Policy Choices
Panelists Kara Kockelman & Lisa Loftus‐Otway, UT Austin

2.
Self-driving cars are game changers.
 Revolution or Evolution....
 Where do you think we are?
 A lot can happen in 13 years…

3.
1900: 5th Ave NYC Easter Parade
Spot
the car!

4.
1913: 5th Ave NYC Easter Parade
Source: George Grantham Bain Collection
Spot the
horse!

5.
Introduction
“You can count on one hand the number of years it will
take before ordinary people can experience (AVs).”
–Sergey Brin, at the 2012 signing of California’s SB 1298.

6.
Opportunities for CAVs
 U.S. Safety
 In 2014, 6.0 million crashes in the U.S. resulting in 32,675 deaths &
>$500+ billion in comprehensive costs.
 Driver error is primary cause of > 90% of U.S. crashes.
 40% of fatal crashes involve alcohol, drugs, fatigue &/or distraction.
 AVs can dramatically impact safety by reducing human errors.
 U.S. Congestion
 7 billion hours of delay & $160 billion losses in 2014.
 Reductions possible via traffic smoothing, tighter headways,
cooperative adaptive cruise control (CACC) & fewer crashes.

7.
Opportunities (2)
 Traveler Behaviors
 Car‐sharing & ride‐sharing
 Increased mobility for
elderly, disabled, & children?
 Parking benefits
 Latent & induced VMT
 Freight Movement
 Reduced labor & thus shipping costs
 Improved fuel economies from tight‐headway drafting
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
Midnight
3:00 AM
6:00 AM
9:00 AM
Noon
3:00 PM
6:00 PM
9:00 PM
Midnight
Vehs ≤ 10 yrs
Vehs ≤ 15 yrs
All vehs

8.
Part 1
What Are AVs Worth?
- to Society & to Individual Owners

9.
Annual Per-AV Economic Impacts
Assumed Market Share
10% 50% 90%
Crashes
Lives Saved 1,100 9,600 21,700
Economic Savings per AV per year $430 $770 $960
Comprehen. Savings per AV per yr $1,390 $2,480 $3,100
Congestion
Travel Time Savings (M Hours) 756 1,680 2,770
Fuel Savings (M Gallons) 102 224 724
Savings per AV per year $1,320 $590 $550
Other Impacts
Parking Savings per AV per year $250 $250 $250
VMT Increase 2.0% 7.5% 9.0%
Change in Total # Vehicles ‐4.7% ‐23.7% ‐42.6%
(U.S. Context, $2012)

10.
Totaled Social Benefits
Assumed Market Share
10% 50% 90%
Annual U.S. Savings: Economic Benefits Only $26 B $102 B $201 B
Annual U.S. Savings: Comprehensive Benefits $38 B $211 B $447 B
Savings Per AV per year: Econ. Benefits Only $2,000 $1,610 $1,760
Savings Per AV per year: Comprehen. Benefits $2,960 $3,320 $3,900
Net Present Value (NPV) of AV Benefits minus
Purchase Price (Econ. Benefits Only) $5,200 $7,250 $10,400
Net Present Value of AV Benefits minus Purchase
Price (Comprehensive Benefits) $12,500
$20,300
per AV sold $26,700
Added Purchase Price $10,000 $5,000 $3,000

11.
U.S. Industry Impacts, at 100% Adoption
Industry
Industry Size
($B/yr)
Industry Impact
($B/yr)
% Change in
Industry
$ per Capita per Year
Insurance $180B/yr $108B/yr 60% $339/person/yr
Freight Transportation $604 $100 17% $313
Land Development $931 $45 5% $142
Automotive $570 $42 7% $132
Personal Transport $86 $27 31% $83
Electronics & Software
Technology
$203 $26 13% $83
Auto Repair $58 $21 36% $66
Digital Media $42 $14 33% $44
Medical $2,700 $12 0% $36
Oil and Gas $284 $10 4% $31
Construction/Infrastructure $169 $8 4% $24
Traffic Police $10 $5 50% $16
Law $277 $3 1% $10
Industry-based Totals $6,113 $420B/year 7% $1,318/person/year

12.
Adding in Additional Effects:
Travel Time “Productivity” Rises
+ Pain & Suffering from Crashes Fall
Economy-Wide (non-Industry-based) Effects
Economic Impact ($B/yr) $ per Capita per Year
Productivity en
Route
$645 Billion/year $2,022/person-year
Pain & Suffering +
other Crash Costs
$488 B/year $1,530
Additional Effects $1,133 B/year $3,552/person-year
Overall Totals
(industry + other)
$1.4 Trillion per year! $4,419 per person-year

13.
Part 2
Forecasting Americans’ Long-Term
Adoption of Connected & Autonomous
Vehicle (C/AV) Technologies

14.
Willingness to Pay (WTP)
Average
WTP
Average WTP
(if WTP > 0)
% of Respondents
with $0 WTP
Electronic Stability Control $52 $79 33.4%
Lane Centering $205 $352 41.7%
Left Turn Assist $119 $221 46.1%
Cross Traffic Sensor $169 $252 32.8%
Adaptive Headlight $203 $345 41.1%
Pedestrian Detect $145 $232 37.5%
Adaptive Cruise Control $126 $202 37.7%
Blind Spot Monitoring $160 $210 23.7%
Traffic Sign Recognition $93 $204 54.4%
Emergency Automatic Braking $183 $257 28.7%
Level 3 Automation $2,438 $5,470 55.4%
Self‐parking Valet System $436 $902 51.7%
Level 4 Automation $5,857 $14,196 58.7%
Connectivity (DSRC) $67 $111 39.1%

15.
Simulating Fleet Evolution
Vehicle inventory
Demographics
Travel Patterns
Technology evolution
Transaction
decision model
(multinomial logit)
Add technologies
to old vehicles
Sell a vehicle
and buy vehicles
Buy vehicles
Sell a vehicle
Add connectivity
if WTP≥ Price
Buy new or
used? (Logit)
LV4 WTP
≥ Price
LV3 WTP
≥ Price
Dispose of
the oldest vehicle
Add connectivity
if WTP≥ Price
Vehicle
is already
LV3 or LV4
End: Do nothing
End: Dispose of
the oldest vehicle
New
Used
End:
Add LV4
End: Add LV1,
LV2, or self-
parking valet
if WTP≥ Price
No
End:
Add LV3
No
No
Yes
Yes
Yes
Same process for each household, every year.

16.
Predicted Shares of US Light-duty Vehicles
59.5%
83.5%
100.0% 100.0%
100.0% 100.0%
100.0%
0.0%
100.0%
0%
25%
50%
75%
100%
2015 2020 2025 2030 2035 2040 2045
DSRC‐based Connectivity
43.0%
43.8%
24.8%
43.4%
43.2%
70.7%
59.7%
0.0%
87.2%
0%
25%
50%
75%
100%
2015 2020 2025 2030 2035 2040 2045
Level 4 Automation

17.
Part 3
Agent-Based Models for Shared AVs
+ = &
 Less than 20% of newer (& 15% of all) personal vehicles are in‐use at
peak times, even with 5‐minute pickup & drop‐off buffers.
 Car‐sharing programs like ZipCar & Car2go have expanded quickly,
with the number of U.S. users doubling every year or two, over the
past decade.
 Shared Autonomous Vehicles (SAVs) can help overcome car‐sharing
barriers, like return‐trip certainty & vehicle access distances.

18.
Agent-Based Model Framework
 Grid‐based 10 mi x 10 mi urban area with 0.25‐sq. mile zones.
 Trip generation:
 Poisson‐based PK & OP counts for trip generation, every 5 minutes.
 Higher trip production & attraction rates closer to city center.
 Mostly round‐trip travel, with 78% travelers returning via SAVs.
 Random departure times & trip distances (2009 NHTS).
 SAVs travel at fixed speeds, with 5 min. intervals.
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
4.5%
5.0%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1 2 3 4 5 6 7 8 9 1
0
1
1
1
2
1
3
1
4
1
5
Midnight ‐ 3 AM
3 AM ‐ 6 AM
6 AM ‐ 9 AM
9 AM ‐ Noon
Noon ‐ 3 PM
3 PM ‐ 6 PM
6 PM ‐ 9 PM
9 PM ‐ Midnight
Trip Generation Trip Distances (mi.)Dwell Times (hrs.)

19.
Example: One SAV’s 24-hour Journey
Higher AM Trip
Attraction
Higher PM Trip
Attraction
• Red Arrows
SAV Relocation
• Blue Arrows
Serving Riders
5 mi
10 mi
SE

20.
Case Study Results
 100 days were simulated to assess SAV travel implications.
Parameter Value
Service area 10 mi. x 10 mi.
Outer trip generation rate 9 trips/cell/day
CBD edge trip generation rate 27 trips/cell/day
CBD core trip generation rate 30 trips/cell/day
Off‐peak speed 33 mph
Peak speed 21 mph
AM peak 7 AM ‐ 8 AM
PM peak 4 PM ‐ 6:30 PM
Trip share returning by SAV 78%
Scenario Results
 Each SAV replaced 9 to 13
conventional vehicles.
 Avg. wait time ≈ 2.8 min.
 11% new/induced (empty‐
vehicle) travel.
 Yet 5% to 50% (GHG vs. VOCs)
life‐cycle emissions reductions,
thanks to smaller vehicles,
fewer cold starts, & less
parking infrastructure!

21.
Part 4
What if SAVs Serve Central Austin, &
Offer Dynamic Ride-Sharing (DRS)?

22.
One SAV’s
24-hour day
Dropoff
Pickup
Travel

23.
Case Study Results
 1:10 & 1:8 veh. replacement
rates (with & w/o DRS)
 System pays for itself with
just $1/mile fares!
 Electric vehicles (Leaf &
Model S) also tested (with
inductive charging), using
100 mi x 100 mi region.
 DRS saves more emissions ‐
& VMT even falls (vs. BAU).
Measure With DRS Without DRS
SAV fleet size 1,855 2,181
Veh. replacement rate 9.95 8.47
Average wait time 57 sec 47 sec
% Waiting > 10 min. 0.60% 0.33%
5‐6 PM avg. wait 3.0 min 2.4 min
Avg. total trip time 14.4 min 13.8 min
New VMT introduced 4.90% 7.92%
# rides shared 5,754 0
% VMT shared 4.50% 0%
 24‐hour days simulated with 56,300 to 270,000 trips served.
 Excellent Level of Service (typ. wait time < 3 min.)

24.
Part 5
What if SAVs Serve the Entire Region?
And Are SAElectricVs?

25.
Shared Autonomous Electric Vehicles
SAEVs
or

26.
Station Generation via 30-day Initial Run
Check for
unmet
requests
Find closest
SAEV
SAEV has
range to meet
trip request?
Unmet
request Send SAEV to
serve trip
Yes
t=t+1sec
Next
timestep:
t=t+1sec
No new
requests
None
available
Is a charging
station also in
range?
No
Create new
station at
vehicle’s
location
No
SAEV begins
charging
SAEV heads
to closest
station
SAEV removed
from
consideration
Try again, in
next timestep
Yes

27.
Charging Station Locations
 Charging stations generated based on demand.
 Number of charging stations formed is dependent only
on vehicle range.
 Stations
formed for
200‐mile
range (left)
& 60‐mile
range
(right)

28.
Central Austin Station Locations
Assuming 60 mile range + 4 hour charge time + 5:1 travelers per
SAEV (= 28 stations over 6 x 10 mi area)

29.
Austin SAEV Results
Scenario
Gas
SAV
Short‐
Range
SAEV
Long‐Range
Fast Charge
Long‐
Range
SAEV
Short‐Range
but Fast
Charge
Long‐Range,
Fast Charge,
Smaller Fleet
Range (mi) 525 60 200 200 60 200
Recharge/Refuel Time (min) 2 240 30 240 30 30
# of Charging/Gas Stations 19 155 155 155 155 155
Fleet Size (# vehicles) 5,893 5,893 5,893 5,893 5,893 4,124
Avg. Daily miles per Vehicle 452 201 354 441 355 501
% of Unserved Trips 1.62 60.6 19.4 2.67 16.2 15.2
Avg. Daily Trips per Vehicle 28.5 11.4 23.4 28.2 24.3 35.1
Avg. Wait Time Per Trip (min) 4.45 9.82 8.76 5.49 6.16 9.55
% Unoccupied Travel 6.05 13.1 7.88 6.86 14.2 8.62
% Travel for Charging 0.65 5.59 1.26 1.05 5.34 1.27
• Fleet size is key to lower response times. Tripling fleet size (from 9:1 to 3:1
travelers per SAEV) lowers average response times by >75%.
• Longer charge times increase response times (& unserved trips rise 19% to 61%)
• Longer ranges lower empty VMT, but fast‐charging improves response times.
• Trips in Austin’s urban core are served best (e.g., never exceed 30‐min wait times).

30.
SAEV Cost Assumptions
• Conventional BEV Costs: $25,000 (short range) to $35,000 (long‐range)
• Self‐driving Technology Cost: $5,000 to $25,000 per vehicle
• Battery Replacement: $100 ‐ $190 per kWh (once per vehicle life)
• Vehicle Maintenance: 5.4¢ to ‐6.6¢ per mile
• Insurance & Registration: $550 ‐ $2,200 per vehicle‐year
• Electricity: 8¢ to 20¢ per kWh
• Level II Chargers: $8,000 ‐ $18,000 each
• Level II Charger Maintenance: $25 ‐ $50 per year, per charger
• Fast (Level III) Charger: $10,000 ‐ $100,000 per charger
• Fast Charger Maintenance: $1,000 ‐ $2,000 per year, per charger
• Station Properties: $1,980 to $6,900 per vehicle space (based on location)

31.
Financial Results: Costs per Mile
Mid‐Range
Expected Costs
per mile
Gasoline
SAV
Short‐
Range
SAEV
Long‐
Range
SAEV
Fast‐
Charge,
Long‐
Range
SAEV
Fast‐
Charge
SAEV
Fast‐Charge,
Long‐Range
Reduced
Fleet
Electricity/Fuel 6.39¢/mi 4.51 4.26 4.21 4.57 4.29
Vehicle Maint.,
Admin +
Attendants
18.4¢/mi 19.7 18.6 18.4 19.9 18.7
Charger Costs (Land
+ Infrastructure)
n/a 3.57 1.35 2.15 6.30 0.76
Vehicle Purchase 19.6¢/mi 27.7 29.4 28.3 25.3 28.4
Battery Costs n/a 1.58 4.91 4.85 1.60 4.95
Total Costs per Mile 45¢/mi 59¢/mi 59¢/mi 59¢/mi 59¢/mi 59¢/mi
Daily Vehicle Profit
($1/mile fare)
$234
/veh‐day
$72 $132 $170 $126 $187
#Trips/vehicle‐day
28 trips
/veh‐day
11 23 28 24 35
Response time/trip 4.4 min 9.8 8.8 5.5 6.2 9.6

32.
Part 6
How Should We Modify our Travel
Demand Models & Plan for the Future?

33.
More Complete Model Assumptions
 Vehicle ownership changes over time (AVs cost more & SAVs
allow people to avoid ownership).
 Travel times are less burdensome (for drivers)
 Lower values of travel time thanks to more productive (& restful!) in‐
vehicle activities
 Affect trip mode choice probabilities
 Travel costs may fall
 AVs can head to lower‐cost parking locations
 Shared AVs reduce overall vehicle‐use costs
 Dynamic ride‐sharing reduces per‐trip costs even further
 Link capacities rise on roadways
 V2V communications (e.g., CACC) + smart intersections (long term)
 AVs may eventually follow at shorter headways & distances
 Hopefully lower travel times & travel‐time unreliability…

34.
However, we also expect…
 Longer travel distances (more distant destinations).
 More trip‐making by those presently unlicensed, with
disabilities &/or other difficulties driving.
 Less air travel by passengers & rail travel by freight.
 Possibly larger, less‐efficient vehicles, for longer‐distance
trips, & more land use sprawl.
This means…
 Rising congestion & infrastructure damage in many locations.
 Need for smarter system management, including incentives
for ride‐sharing & non‐motorized travel, route guidance,
credit‐based congestion pricing & micro‐tolling ‐ to
internalize externalities & operate more efficiently, equitably,
& sustainably!

35.
In Conclusion…
 CAVs offer tremendous benefits for mobility, safety & parking,
but will add VMT & congestion.
 SAVs offer a new & exciting (transit?) mode, with each SAV
replacing ~8 personal vehicles, for same level of motorized trip‐
making.
 SAVs add 7‐10% extra VMT (though DRS may reduce VMT).
 Yet SAVs may bring useful travel‐cost savings, emissions benefits
+ profits for transit providers.
 Traditional travel models cannot capture the details of SAV
systems & CAV operations. Microsimulation is needed.
 Smart system management practices are also needed, to avoid
gridlock, sprawl, greater energy use, & other downsides.

36.
So, what is going on in ‘the law’
around the world?

37.
United States: Federal
• National Highway Traffic Safety Administration (NHTSA)
preliminary policy on Automated Vehicles in 2013
– Outlined definitions for Levels 0 through 4 of automation
• NHTSA in September 2016 issued new Policy on Autonomous
Vehicles
– Adopted SAE J3016 definitions (L0 through Level 5) as
their standard.
• Deliberately issued as policy & not regulations, with goal to
set stage for consistent national framework but providing
flexibility to states.

38.
NHTSA 2016 Policy
Delineated roles/responsibilities for state(s) policy:
‐ States retain their traditional responsibilities for vehicle
licensing, registration, traffic laws & enforcement, and motor
vehicle insurance & liability.
‐ NHTSA continued preemption for interpretations,
exemptions, notice, and rulemaking & enforcement
authority.
Manufacturer responsibility to determine their system
conforms with SAE J3016.

39.
NHTSA’s Framework for
Vehicle Performance Guidance

40.
U.S.: Federal (2)
• NHTSA October 2016: Policy on Cyber Security in
Autonomous Vehicles
– Covers all vehicles not just HAVs & applies to designers,
supplies, manufactures & modifiers
• FAST Act 2015, §24302 limitations on data retrieved from
Event Data Recorders (EDRs)
– NHTSA required to determine amount of time EDRs
should capture & record data for retrieval ... to provide
sufficient information to investigate a motor vehicle
crash

41.
U.S.: State‐level
• Over 80 bills are currently in front of U.S. state legislatures on
this topic.
• Nevada created legislation allowing testing in 2011.
• California legislation authorized a pilot program in 2014.
• Michigan: 2013 allowed testing of automated vehicles as long
as human was in car.
• Michigan: 2016 allows driverless cars to be driven for any of the
following purposes, no human required to be in car:
– Personal use; road testing; as part of a SAVE program or “on‐
demand automated vehicle network;” & as part of a platoon.

42.
California’s & Michigan’s approaches differ...
 California: prescriptive approach
 Required rulemaking by agencies, pilot tests must be
authorized, test vehicles do not require driver behind
wheel, but qualified test drivers must have ability to take
control, minimum insurance surety bond of $5 million.
 Michigan: framework approach
 Initially, Automakers can test AVs as long as human in car
in original legislation. No agency rulemaking required.
 Current legislation, allows driverless cars to be driven for
multiple purposes not just road testing, without a human
in the vehicle as the AI is considered the driver.

43.
Stats from California’s pilot.....
• 17 testing permits (mainly Tier 1’s, OEMS & technology
manufacturers)
• 26 traffic accidents involving HAVs (Google 22, Delphi 1, Cruise
1, GMCruise 1, Nissan 1)
• Reported disengagements from automated mode between Dec
1, 2015 & Nov 30, 2016:
– Bosch – 1442 – 982 miles driven(MD)
– BMW – 1 – 638 MD
– Delphi – 178 – 3125 MD
– Ford – 3 – 590 MD
– Google – 465 – 1,060,199 MD
– GM Cruise – 284 – 9970 MD
– Nissan – 28 – 4,099 MD
– Mercedes – 336 – 673 MD
– Tesla – 182 – 530 MD
https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/autonomousveh_ol316
https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/disengagement_report_2016

44.
European Union
EU has not yet passed any legislation specifically on HAVs.
 EU Directive 2007/46 EEC regulates how vehicles should be designed
& operated.
 EU Roadworthiness Directive 2014/45 EU sets out basis for
roadworthiness.
Problems with the Vienna Convention on Road Traffic Article 8’s
language stating ‘every driver shall at all times be able to control his
vehicle’ incidentally prevented the development & testing of HAVs.
 Amendments to the Convention in 2016 allow drivers to take hands
off wheel in self‐driving cars.
ECE Regulation 79 also creates an impediment through requirements
for specific steering configurations:
 advanced driver steering system is only allowed to control steering as
long as the driver remains in primary control of vehicle at all times.

45.
EU Member States
 Finland, France, Germany, Netherlands, Sweden & UK
implemented legislation in 2015/2016
 Finland, Netherlands & Sweden, all have similar
systems to California’s legislation & regulations for
pilot tests, & for driver licensing.
 UK issued code of practice in July 2015, which must be
followed by any groups conducting HAV testing. This
includes licensing & training provisions, & a risk
management process by the testing group.
 UK issued Vehicle Technology and Aviation Bill 2016‐
2017 in February 2017 – outlines liability for insurers of
automated vehicles.

46.
Activities in Canada, Japan & Australia
 Canada has not yet created federal regulation
 Province of Ontario in 2016 produced legislation &
regulations for a pilot program
 Japan has allowed road testing, & is working to
develop regulations
 Australia has not federally legislated
 National Transport Commission has set out
recommendations & policy positions in 2016
 Government of New South Wales introduced legislation in
September 2015 for road testing. NSW released a future
transport roadmap in fall 2016 that outlined the ministry’s
view on the transition to HAVs

47.
Initial Legal Issues
 Privacy, Liability, Cybersecurity & Freedom of
Information Requests / State Open Records
Requests all raised as concerns for automated &
connected vehicles.
 No case law yet on these issues.
 NHTSA & FTC have noted they are reviewing
hacking & privacy of consumer data in HAVs.
 Federal statutes also provide penalties under the
Computer Fraud and Abuse Act, Digital Millennium
Copyright Act, Wiretap Act, & Patriot Act.

48.
Initial legal issues (2)
 Privacy realm three areas have been identified
as needing changes to law:
1. Autonomy privacy (i.e. an individual’s privacy
under 4th amendment to the U.S. Constitution
e.g. illegal search & seizure);
2. Personal information privacy, and
3. Surveillance.
• California passed law regarding consumer
privacy in HAVs

49.
Initial Conclusions
 Many jurisdictions around the world have
begun to draft legislation & regulations with
a primary focus on pilot testing.
 California has begun to review how it needs to
amend its laws & regulations.
 Legal articles have primarily focused on
privacy, liability, cyber security &
constitutional protections.

50.
Thank you!
Questions & Suggestions?
kkockelm@mail.utexas.edu
loftusotway@mail.utexas.edu