Login

Project

#98 Risk, Liability and Insurance framework for Autonomous Vehicles


Principal Investigator
Rahul Mangharam
Status
Completed
Start Date
July 1, 2017
End Date
June 30, 2018
Project Type
Research Applied
Grant Program
MAP-21 TSET National (2013 - 2018)
Grant Cycle
TSET - University of Pennsylvania
Visibility
Public
Photo:
Download

Abstract

The prevailing approach to liability and insurance for automobile accidents is based on a formal model of individual driver responsibility that operates within a practical reality of collective responsibility.  Except in the case of “no fault” automobile insurance systems (which never truly eliminated fault and are dwindling in importance from a national perspective), all automobile liability payments are predicated on driver fault.  Although the entitlement to compensation is predicated on individual driver fault, the compensation comes from automobile liability insurance that is priced based on demographic and other factors that are only weakly correlated with the safety and competence of individual drivers. This project will investigate the technical toolchain to compute the risk of autonomous vehicles within an urban environment and the regulatory models for insurance of vehicles with perception, planning and decision making control. 
    
Description
A large-scale shift to AV has the potential to fundamentally alter this prevailing approach. If we retain a fault-based liability system (the most likely scenario in the short to medium term), then the locus of responsibility will increasingly shift from individual humans to the designers of the AV systems and the manufacturers of the vehicles in which those systems are housed. That is a profound shift along multiple dimensions, each of which will be addressed in the research:
1.	Identifying the gap between driver-based and AV liability: At present, the liability of any individual driver is, as a practical matter, capped at the limit of his or her automobile liability insurance, notwithstanding the fact that the legal liability of the driver might be much higher, because of asset protection laws and practices (e.g. protected retirement savings) that have made the U.S. into what legal scholars call a “judgment proof society.”  This significantly limits the total payments in automobile liability cases.  These asset protection laws and practices do not apply to corporations.  In addition, there are returns to scale from litigating against a single AV designer or manufacturer that may further expand the total liability payments from automobile accidents even if AV systems reduce the total number of accidents.  The research will include working with available automobile liability insurance data to estimate the gap between accident costs and liability payments attributable to these caps and developing models to predict how these changes in the locus of responsibility may increase the payments on both an individual accident and aggregate basis. 

We will use the 6 Million cases in the 2004 General Estimates System (GES) crash database from the NHSTA Pre-Crash Scenario Typology for Crash Avoidance Research report. This will be used to define and statistically describes new pre-crash scenario typologies for light vehicles. Each pre-crash scenario depicts vehicle movements and dynamics as well as the critical event occurring immediately prior to a crash. The goal of this typology is to establish a common AV safety benchmark, which will allow researchers to determine which traffic safety issues should be of first priority to investigate and to develop concomitant crash avoidance systems.

2.	Evidence-based Product Liability for AVs: At present, liability for automobile accidents is based on individual driver negligence, with the liability of manufacturers and other system designers (roads, traffic signals, etc.) of relatively marginal importance, as illustrated by the fact that aggregate U.S. automobile liability insurance premiums are many times larger than aggregate U.S. product liability insurance premiums (for all products, not just autos).  The liability of AV systems and automobile manufacturers will be based on product liability law, not standard negligence law.  The research will include a qualitative investigation of the differences in standards and burdens of proof in a product liability regime as compared to an individual driver responsibility regime, and the potential relationship between the development of ex ante testing and certification requirements involved in “A Driver’s License Test for Autonomous Vehicles” and ex post product liability.

3.	Relating AV Risk and Liability: At present, automobile liability insurance premiums are based on backward looking actuarial models that link accident frequency and severity to demographic and vehicle-related variables.  Liability insurance for AV designers and auto manufacturers will be more like product liability insurance than individual auto liability insurance.  Because product liability insurance has never been tested in the automobile context with the volume of claims that may be presented in the AV era, backward looking actuarial models will not be sufficient.  The research will include developing models predicting the frequency and severity of AV accidents and translating those models into forms that are useful for insurance underwriting.

4.	Pricing AV Liability: At present, the vehicle-related pricing differences for automobile liability insurance (e.g. lower premiums for mini vans than sports cars) largely reflect the association between vehicle types and the riskiness of the people who choose to buy the vehicles, not the safety of the vehicle.  As vehicles become more autonomous, the liability insurance premium associated with the vehicle will increasingly reflect the safety of the (increasingly autonomous) vehicle.  In addition, because the liability belongs to the manufacturer and suppliers and not the individual drivers, the liability insurance premium will become part of the price of the car.  The research will include modeling of the potential impact on the price of personal automobiles and consideration of how that pricing may affect consumer behavior, including an analysis of the potential that this internalization of the costs of accidents will provide an incentive for safety.  

An additional qualitative component of the liability and insurance research will be translating the research findings and recommendations of the engineering team into concepts and language that make sense to the major players in the liability and insurance field.  
Timeline
July 1, 2017 to June 30, 2018
Strategic Description / RD&T

    
Deployment Plan
We will use the 6 Million cases in the 2004 General Estimates System (GES) crash database from the NHSTA Pre-Crash Scenario Typology for Crash Avoidance Research report. This will be used to define and statistically describes new pre-crash scenario typologies for light vehicles. Each pre-crash scenario depicts vehicle movements and dynamics as well as the critical event occurring immediately prior to a crash. The goal of this typology is to establish a common AV safety benchmark, which will allow researchers to determine which traffic safety issues should be of first priority to investigate and to develop concomitant crash avoidance systems.
Expected Outcomes/Impacts
An additional qualitative component of the liability and insurance research will be translating the research findings and recommendations of the engineering team into concepts and language that make sense to the major players in the liability and insurance field.  
Expected Outputs

    
TRID


    

Individuals Involved

Email Name Affiliation Role Position
rahulm@seas.upenn.edu Mangharam, Rahul University of Pennsylvania PI Other
alena.rodionova@seas.upenn.edu Rodionova, Alena University of Pennsylvania Other Student - PhD

Budget

Amount of UTC Funds Awarded
$0.00
Total Project Budget (from all funding sources)
$

Documents

Type Name Uploaded
Publication MOBILITY21-Strategic-Investments-for-Transportation-Infrastructure-Technology.pdf April 17, 2018, 12:27 p.m.
Presentation Mangharam_-_Drivers_License_Test_for_Driveless_Vehicles.pdf April 17, 2018, 12:27 p.m.
Progress Report 98_Progress_Report_2018-03-31 April 17, 2018, 12:27 p.m.
Publication Computer_Aided_Design_for_Safe_Autonomous_Vehicles_f1cmTo5.pdf Nov. 30, 2018, 9:21 p.m.
Publication AVs_-_Safe_at_Any_Speed.pdf Nov. 30, 2018, 9:21 p.m.
Progress Report 98_Progress_Report_2018-09-30 Nov. 30, 2018, 9:21 p.m.
Final Report 98_-_AVs_-_Safe_at_Any_Speed.pdf March 26, 2019, 6:12 a.m.

Match Sources

No match sources!

Partners

No partners!