Casualty Actuarial Society - cdn.ymaws.com

Casualty Actuarial Society

Automated Vehicle Task Force (CAS AVTF)

Wenwen Salerno, ACAS, MAAA

August, 2015

Vehicle operates as Windows System

•  Automobiles would frequently crash for no apparent reason. This would be so common that motorists would simply accept it, restart their car and continue driving.

•  Airbags would ask, 'Are you sure?' before deployment. •  Vehicles would occasionally shut down completely and

refuse to restart, requiring motorists to reinstall their engine.

2 www.vtpi.org/avip.pdf

3

Example

  Video

CAS AVTF: Overview

4

Pre market Post market Post claim

identify & quantify risks

accurately price the technology

compensate claimants fairly & efficiently

Focus

The CAS AVTF is researching the technology’s risks to provide policymakers with the information needed to ensure the product is brought to market as safely and efficiently as possible.

Goal

Agenda

  Automated Vehicles – Background

  Insurance Industry Impact

  Automated Vehicle Risk Profile

  Vehicle Symbol Analysis

  Regulatory Overview

5

Agenda






6

Automated Vehicle Components

  Automatic transmissions   Diverse sensors (cameras, radars, etc.)   Wireless network   Navigation system including GPS   Automated control (Steering, braking, signals, etc.)

  Servers, software and power supply

7 www.vtpi.org/avip.pdf

Driver Assistance Features

8

Three key technologies for AVs

9

V2V/V2I

LiDAR

Inertial Navigation

Systems & GPS

•  Vehicle to Vehicle or Vehicle to Infrastructure •  Dedicated Short Range Communications (DSRC), similar to wifi •  Vehicle to communication to other vehicles or infrastructure (traffic

signals, toll booths, etc

•  Light Detection And Ranging •  combination of light and radar, and uses laser light to create 3D images

of the surrounding environment. •  Remote sensing technology to measure distances

•  INS uses computers, accelerometers (motion), and gyroscopes (rotation) •  Calculates position, orientation, and velocity

10

Cameras and Deep Learning may be another technology NVIDIA @ CES 2015

  NVIDIA video   https://www.youtube.com/watch?v=o29TBy2a0ek

Historic Developments

11

2005 – 2010

• 05: Stanford wins DARPA Grand Challenge

• 07: CMU wins DARPA Urban Challenge

• 09: Google begins tesCng on public roads

• 09: EU launches Project SARTRE

• 10: Volvo City Safe standard

2011

• Google surpasses 150K miles

• BMW begins tesCng self driving car on public roads

• NV passes autonomous car law

2012

• Google @ 300K accident free miles

• Nissan opens research facility in Silicon Valley

• Google & ConCnental receive autonomous car licenses

• FL & CA pass autonomous car laws

2013

• Google @ 500K miles

• Oxford creates a $7,750 self-‐driving system

• Public road tesCng • Britain • Mercedes • CMU

• Audi receives AV car license

• NHTSA issues policy on AVs

• DC passes AV car law

2014

• MI passes AV law

• NHTSA passes V2V

• Google @ 700k miles

• Volvo ‘Drive Me’ tests in Gothenburg

• Google developing driverless car without steering wheel or brakes

2015 Developments

12

January •  Audi partner with Nvidea to use Tegra X1 chips •  Nissan & NASA announce automated research project partnership

February •  Uber and Carnegie Mellon University announce strategic partnership •  Apple reported to be working on automated vehicles Sony’s image sensors and ZMP’s

robotics

April •  Delphi 3,400 mile cross country trip with 99% of miles in autonomous mode •  Nevada licenses Mercedes’ Freightliner truck •  Nokia’s HERE: 2 billion offer from auto manufacture and Uber reported to offer $3 billion

May •  Tesla adjusts adaptive cruise control to limit liability •  Daimler’s self-driving truck got license and hit the road July •  M City in University of Michigan

London’s AV testing regulations may increase investment

13

Goal •  Light-touch non-regulatory approach •  provides clarity for industry to invest in further

in research

No special licenses or

permits requires

No geographical limits

No additional insurance

requirements

London has 3 trials underway

14

Milton Keyes and Coventry Greenwich

•  Lutz Pods that drive in pedestrian zones

•  Max speed 15 mph •  Electronic AV

UK Autodrive Programme: 3 years to pave way for introduc?on of AVs Dept. of Transporta.on put ~$29M USD for trials

Explore both legal and technical changes required for Autonomous Vehicles

Bristol

•  GATEway shucles •  Electronic AV •  Local tour with drop

off points: input desCnaCon on CPU

•  Venturer consorCum will invesCgate congesCon and safety

•  BAE Wildcat

Future development may create three models for AVs

Private: all driving/limited loca?on Private: Some driving/all loca?ons

•  Takes over some of the driving •  E.g. Supercruise, parallel parking •  Only operates in specified area •  Driver owns and operates •  Mercedes, BMW, Volvo, Cadillac,

Telsa

•  End to end service •  Only operates in specified area •  “Taxi” service •  Google, Uber

16

Commercial: Some driving/Special Vehicle

http://www.bloomberg.com/news/articles/2015-05-14/daimler-s-freightliner-tests-self-driving-truck-in-nevada

  Daimler’s video

17

Commercial: Some driving/Special Vehicle

Agenda






18

Private Passenger Auto Insurance Coverage Components

19

Liability -‐ Bodily Injury Liability, $131

Liability -‐ Property Damage Liability, $78

Liability -‐ Medical Payment, $7

Liability -‐ Personal Injury Protec?on, $104

Liability -‐ UM/UIM Bodily Injury, $85

Physical Damage -‐ Comprehensive, $50

Physical Damage -‐ Collision, $105

Other -‐ Addi?onal Service, $17

PRIVATE PASSENGER AUTO INSURANCE EXAMPLE

From presenter’s example quotes

Physical Damage/Liability Coverages

20

Physical Damage/Liability Coverages

21

Liability Coverages

22

Private Passenger Auto Liability vs. Products Liability

23

  ALAE Factor

  Permissible Loss Ratio

  Accident classification

  Liability limits

  Settlement lag

  Unnecessary coverages

Insurers’ responses

24

Pricing Adjustments   Proprietary coverage level vehicle symbols

Paten?ng Pricing Approaches   State Farm: Trip-‐Based Insurance Pricing Plan (2015)   Travelers: “Risk-‐Zone” Pricing (2014)   Progressive: Vehicle sensor approach (2012)

Forming Partnerships   Ford, State Farm & U of Michigan – Ford Hybrid automated research vehicle (Dec 2013)   Honda & major insurance company sign agreement to use self-‐driving automobile test track

at former Concord Naval Weapons StaCon (March 2015)

Tes?fying at hearings   CA DOI: State Farm & NaConwide & CAS AVTF   NJ Senate: Munich Re America

Industry Groups are performing research   CAS; HLDI-‐IIHS; RAND Corp; Brookings InsCtute

Agenda






25

“93% of accidents are caused by human error.”

26

National Motor Vehicle Crash Causation Survey (NMVCCS) – Limiting Factors

27

0%

10%

20%

30%

40%

50%

1 2 3 4 1 2 3 4 5 6

12.2% 11.6%

0.4%

21.3%

3.1%

11.0%

2.3% 2.9%

16.7%

32.4%

48.9% Technology Issues

Behavioral (Driver) Issues

NMVCCS – Implications of the CAS Study

28

New benchmark should be calculated

• Data is old and unrepresentaCve • Human driving risks <> automated vehicle risks

Appropriate test for each risk • Computer simulaCons for technology’s error rate • SimulaCons provide licle insight into driver’s actual use of technology.

Policy changes can increase AV’s safety • 1% reducCon in accidents is ~55k fewer accidents and $1.4 billion of economic value per year

• Policy cost benefit analysis • E.g. driver training program, automated vehicle only lanes, allowing the Avs to speed

Agenda






29

Vehicle symbol analysis approach

30

Vehicle experience groups

Complements to credibility

Automated vehicle symbol:

op?on 1

•  Each group’s experience is weighted and combined with similar vehicles

•  Vehicle’s body style factor •  Prior year factor

•  Assume a brand new vehicle •  e.g. Mercedes introduces a new fully automated vehicle •  No iniCal prior year factor, growth trend impacts credibility


op?on 2

•  Assume update to a current vehicle •  e.g. all new Honda Civics sold with AV equipment

Vehicle symbol analysis approach

31

Vehicle experience groups

Complements to credibility


op?on 1

•  Each group’s experience is weighted and combined with similar vehicles

•  Vehicle’s body style factor •  Prior year factor

•  Assume a brand new vehicle •  e.g. Mercedes introduces a new fully automated vehicle •  No iniCal prior year factor, growth trend impacts credibility


op?on 2


Vehicle Symbol Calculation

32


op?on 1

•  Assume a brand new vehicle •  e.g. Mercedes introduces a new fully automated vehicle •  No iniCal prior year factor, growth trend impacts

credibility

Vehicle Symbol Discount

Number of Exposures

Loss Aeenua?on

Year 0% 25% 50% 75% 100%

2,500 1 0.0% 0.5% 0.9% 1.3% 1.8%

5,000 2 0.0% 1.4% 2.6% 3.9% 5.1%

7,500 3 0.0% 2.8% 5.1% 7.4% 9.7%

10,000 4 0.0% 4.4% 8.0% 11.6% 15.2%

Vehicle Symbol Calculation

33

Vehicle Symbol Discount Loss Aeenua?on

Year 0% 25% 50% 75% 100%

1 0.0% 4.3% 7.4% 10.5% 13.6%

2 0.0% 7.1% 13.7% 20.0% 26.3%

3 0.0% 9.7% 18.2% 25.7% 35.4%

4 0.0% 11.1% 21.0% 31.0% 41.2%


op?on 2


Upcoming Projects: Average Accident Rate

Comparable Driver   Match location & type   Match driver characteristics

Issues with NHTSA Data   Only include police reported accidents   Cannot segment by driver type

Insurance Data   Calculate frequencies for different driving segments   Can more accurately define “good driver”   GLM’s lead to a more stable & accurate calculations

34

Upcoming Projects: Auto Liability vs. Products Liability

Quantify the change in costs from liability systems

  Scenario #1: Assume no change in accidents

  Scenario #2: Accident frequency is reduced by X% –  Determine what X needs to equal for Scenario 1 = Scenario 2

  Scenario #3: Cap liability to $Z –  Determine what Z needs to be for Scenario 1 = Scenario 3

  Scenario #4: Combination of Scenario #2 & #3

35

Agenda






36

37

Current U.S. regulatory approach varies by state

Comments §  CA, DC, FL, MI, NV have

regulations that permit operation/testing of AVs

§  May 2013 NHTSA publication

§  Statement with guidance to states on AV regulations

§  Statement outlined NHTSA plans for testing AV technology

hcp://cyberlaw.stanford.edu/wiki/index.php/Automated_Driving:_LegislaCve_and_Regulatory_AcCon

Consumer Protection

38

NHTSA

States

Individuals

•  establishes regulaCons that manufacturers must self-‐cerCfy with

•  State regulators can impose addiConal requirements

•  Individuals can sue manufacturers if an error occurs

Manufactures •  An alliance of twelve automakers will

create a center for sharing informaCon and analysis.

Regulatory approach needs to be updated

39

Insufficient protec?on for consumers and manufacturers •  AlliedSignal, Inc v. Moran

NHTSA does not have the scale: annual budget 60,000

82822

$M IIHS-HLDI

Auto Insurance NHTSA

•  NHTSA lacks financial incenCve of auto insurance industry •  Accurate evaluaCon of risk

Claims handling incen?ves •  CompensaCng claimants fairly and efficiently is a core

competence of the insurance industry •  Auto industry’s competence is not insurance

Questions and Discussion

Casualty Actuarial Society - cdn.ymaws.com

Documents

Transcript of Casualty Actuarial Society - cdn.ymaws.com