-1-

PROPOSAL TITLE: APPLICATION OF AUTONOMOUS COLLISION AVOIDANCE

TECHNOLOGY TO TRANSIT BUSES DRAFT Version 131121 1001

KEY WORDS

Autonomous braking, autonomous driving, bus transit, collision avoidance, connected vehicle, casualty

and liability expense, bus collisions

ABSTRACT

(to come)

PARTICIPATING ENTITIES

Princeton University (Lead)

Washington State Transit Insurance Pool

American Public Transportation Association

Price Waterhouse

ENDORSING/SUPPORTING ENTITIES

(list other insurance pools and individual transit aency members, and other supporting officials)

THE PROBLEM

Bus transit is one of the safest ways to travel. As shown in Figure 1, over the ten year period spanning

2002 to 2011, there has been a significant decline in the number of injuries per year. The rate of injuries

per million passenger miles also has significantly decreased, as shown in Figure 2. However, these

statistics mask a disturbing trend, a significant increase in casualty and liability expense for the bus transit

industry. Figure 2 shows that the total expense for casualty and liability claims has sharply and steadily

increased, at an average rate of 2.8% per year. Nationwide, in 2011 alone, bus transit properties reported

3,260 collisions, 12,928 injuries, and 92 fatalities, resulting in reported casualty and liability expense

totaling $483,076,010. That amounts to an annual average of $8,069 per bus. Over the ten year period

from 2002 to 2011, the bus transit industry, which is largely self-insured, has spent nearly $4.2 billion on

casualty and liability expenses, an enormous drain on the industry’s resources.

-2-

12,000

12,500

13,000

13,500

14,000

14,500

15,000

15,500

16,000

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

An

nu

al R

ep

ort

ed

Inju

rie

s

Figure 1 US Bus Transit Industry Annual Reported Injuries 2002-2011

Source: Federal Transit Administration National Transit Database

0

0.2

0.4

0.6

0.8

1

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 2 Injuries per Million Bus Passenger Miles Source: Federal Transit Administration National Transit Database

-3-

THE OBJECTIVES

This research project will provide answers to the following questions:

Although the absolute number of bus related injuries and the rate of injuries per million passenger

miles have trended downward, why have casualty and liability expenses trended upward?

How can the industry more accurately account for the total costs of bus collisions?

What is the correlation between costs and type of collision?

What types of autonomous collision avoidance technologies have the greatest potential to reduce

casualty and liability expenses?

What requirements and standards are needed to allow transit operators to specify autonomous

collision avoidance technology for new bus procurements and retrofits?

What is the process for testing and certifying autonomous collision avoidance systems?

Given that buses may remain in service for 15 to 18 years, and digital technology can become

obsolete within 18 months to two years, how should buses be configured to allow rapid and cost-

effective replacement of components and retrofit with updated technology?

UNDERSTANDING THE REASONS FOR THE SHARP INCREASE IN CASUALTY AND

LIABILITY EXPENSE

Transit agencies in the United States receiving Federal funds are required to submit a variety of data on

safety and security including reports on incidents, collisions, property damage costs, injuries, and

fatalities in the Safety and Security Module of the National Transit Database (NTD). They also are

required to submit annual data on operating expenses including casualty and liability expenses in the

350

370

390

410

430

450

470

490

510

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

An

nu

al C

asu

alty

an

d L

iab

ility

Ex

pe

nse

($

mill

ion

s)

Figure 3 US Bus Transit Industry Reported Casualty and Liability Expense 2002-2011 Average Increase 2.8% per Year

Source: Federal Transit Administration National Transit Database

-4-

Financial Module of the NTD. Although property damage expenses exceeding certain thresholds are

reported for each incident, the costs of individual casualty and liability claims are not reported by

incident. Reporting instructions indicate that the following costs are to be reported as annual totals:

Physical damage insurance premiums

Recovery of physical damage losses for public liability and property damage insurance premiums

Insured and uninsured public liability and property damage settlement pay outs and recoveries,

Other corporate insurance premiums (e.g., fidelity bonds, business records insurance)

However, many other costs may be reported in other categories and not properly attributed as costs of

collisions. Such costs may include the following:

Workers compensation

In house legal services

Passenger and service delays

Emergency services response

Lost fare revenue

D & A testing

Overtime

Sick time

Accident investigation

Vehicle recovery

In-house vehicle repair

Hearings and discipline

Spare vehicles and replacements

This project will assemble and accurately account for all costs attributable to specific incidents by type of

collision, so that we can more accurately gauge the benefits realizable through adopting autonomous

collision avoidance technology.

THE POTENTIAL TO AVOID COLLISIONS

In 2004, 2007, and 2013 the Federal Transit Administration released research reports on the potential for

technology to reduce bus crashes. i,ii,iii

These reports concluded that the systems envisioned at the time,

which would warn bus drivers of impending collisions, could be effective in reducing costs.

Due to rapid advances in automated driving technology, however, automobiles are now entering the

market with automated systems that not only sense that a collision is imminent, but actually take control

of throttle, steering and braking functions to avoid or mitigate collisions. In particular, systems for autos

are available that will initiate autonomous braking to bring a vehicle to a stop prior to collision with a

pedestrian or vehicle, even if the vehicle operator does not apply the brakes manually. The Highway Loss

Data Institute has concluded that forward collision avoidance technologies have reduced claims.

Collision Avoidance Systems for cars typically are bundled into optional packages that may include:

-5-

Adaptive cruise control

Adaptive headlights

Autonomous emergency braking

Blind spot detection

Cross traffic alert/avoidance

Driver fatigue/inattention alert

Forward collision avoidance

Lane departure warning

Lane keeping assist

Parking assist

Pedestrian detection/avoidance

Rear collision warning/mitigation

Self-parking

Side impact detection

Optional packages vary in price, but most available multi-function packages cost less than four thousand

dollars.

Given the recent advances in automotive collision avoidance systems, their moderate costs, and the

increasing trend in bus casualty and liability claims, the time is ripe to revisit the application of collision

avoidance technology to transit buses. Considering the $8,069 average casualty and liability expense cost

per bus in 2011 alone, it is clear that a business case can be made that autonomous collision avoidance

technology has the potential to significantly reduce overall transit operating expenses.

To that end, we propose a research program with four phases:

1. Create a stakeholder working group of transit agencies, bus manufacturers, technology developers,

and researchers.

2. Conduct a research assessment of why casualty and liability claims are increasing and determine

the potential for automated collision avoidance systems to reduce fatalities, injuries and claims.

3. Initiate the development of functional requirements and standards to allow installation of collision

avoidance and driver assist technology on new transit buses and retrofit of existing buses.

4. Develop a prototype test bed that would allow developers of innovative collision avoidance and

driver assist technologies to work with transit agencies and researchers to expedite development

and deployment.

These phases are described more fully in the section on Work Program.

HOW THIS PROPOSAL COMPLEMENTS RELATED USDOT RESEARCH

The FTA has undertaken several studies of applying collision avoidance technology to buses, but the state

of the art when these studies were conducted had not reached the current level of autonomous control that

is commercially available in the light vehicle market. In addition, these studies did not include an analysis

of the cost of casualty and liability expenses attributable to collisions.

-6-

FTA is sponsoring a Transit Retrofit Package (TRP) project which includes the use of connected vehicle

technology to provide warnings to bus operators of specific hazards, but the TRP does not include

autonomous control. Autonomous emergency braking has been demonstrated by Volvo for large trucks in

Europe, aspects of which may be transferrable to buses.

The research we propose would include tracking the total costs for a sample of collisions,

developing a methodology for estimating the full costs of collisions for each agency, and

determining the reasons for dramatic increases in industry-wide casualty and liability expenses.

There has not been a current and comprehensive look at the costs to agencies of bus crashes. Although

the FTA NTD requires agencies to submit a report on each collision in the Safety and Security reporting

module, they include only an estimate of property damage that exceeds a certain threshold. FTA does not

require reporting of individual casualty and liability claims and many other costs for each incident. FTA

research reports on collision avoidance technology from 2004, 2007, and 2013, all indicated the potential

for positive benefits. However, they had limited data with which to estimate costs. Due to the lack of

accurate estimates for the costs of collisions, it is difficult to assess the true benefit/cost of collision

avoidance technology. In addition, although industry-wide, the annual rate of injuries per million bus

passenger miles and the absolute number of injuries per year have been trending downward, reported

casualty and liability expense has been trending upward.

This proposal focuses on autonomous reaction of the vehicle to an impending collision rather than

providing warnings to the driver.

Autonomous collision avoidance systems that control steering and braking are now available on autos,

and are being tested on trucks, but autonomous technology for buses has not been addressed. A couple of

commercial collision warning products are available that can be used on buses: Clever Devices offers a

system that provides blind-spot warnings to drivers and pedestrians when a bus is turning, and Mobileye

offers a device that warns of impending forward collisions with pedestrians and vehicles. FTA is

sponsoring research and pilot projects for collision avoidance for buses under the USDOT ITS

program. These provide warnings to drivers, and stress the use of connected vehicle technology. While

warning systems are useful, driver surveys showed they could add to driver distraction. Drivers may not

react quickly enough or appropriately to avoid a crash, or may ignore the warning. Autonomous braking

has the potential to be more effective.

The research we propose would develop standardized interfaces allowing multiple vendors to

provide modular and updated systems and components.

Development of standards for collision avoidance technology has not yet been addressed. Some bus

manufacturers indicated they are working on offering collision avoidance systems. We seek to address

the rapid development (and potential rapid rate of obsolescence) of the technology. According to FTA,

the minimum expected life of a transit bus is 12 years although buses can typically remain in service for

15 to 18 years. Digital technology, however, typically becomes obsolete within months or a few years at

most. Proprietary collision avoidance systems acquired with a new bus will likely prove difficult to

support with replacement parts and software during the life of the bus. If we can standardize and provide

specifications for performance, electromechanical interfaces, networking, and sensor locations,

-7-

maintenance and change-out of components would require less time and expense. The potential for

multiple sources should stimulate innovation and help reduce costs.

URGENCY/PRIORITY

This research is a very high priority. Based on 2011 data, an average of 250 people are reported injured

or killed each week by the US bus transit industry, and the average casualty and liability expense totals

$9.3 million per week for the industry. Buses being procured now are expected to last from 12 to 15

years or more. If they are not capable of being retrofitted with collision avoidance technology at

reasonable cost, it will be many years before the rising trend in casualty and liability expenses can be

reversed.

USER COMMUNITY

The user community for this research primarily would be all bus transit operators in North America, but

the results potentially could be applied world-wide.

IMPLEMENTATION

The research would result in a better understanding of the factors driving up the cost of casualty and

liability expenses. Risk management and safety staff at transit agencies will be able to put that knowledge

to immediate use for driver training and acquisition of new technology.

Buses typically have a service life of 12 to 15 years or more, while computing technology becomes

obsolete within a much shorter period, typically 18 months to two years. Consequently, it is expected that

collision avoidance technology will continue to improve over time and costs will drop.

The research would create a set of consensus standards for electromechanical and electronic interfaces

that could be applied to new buses so they can be retrofitted with collision avoidance technology to keep

pace with innovative developments.

The project also would develop a prototype test bed that would allow developers of innovative collision

avoidance and driver assist technologies to work with transit agencies and researchers to expedite

development and deployment.

EFFECTIVENESS

Studies by the Highway Loss Data Institute have shown that collision avoidance technology, especially

forward collision avoidance systems that incorporate autonomous braking, have reduced crashes and

claims for autos. This research will confirm the applicability of the technology to buses, provide a basis

for estimating system installation costs and the potential for collision and claims reduction, and provide

standards to allow economical retrofit as the technology continues to improve.

WORK PLAN

-8-

The proposed work plan is divided into four phases that feed logically, from describing the problem

through data collection and analysis to the development and testing of alternative solutions.

1. Create management and oversight structure

a. Organization/Project Mobilization

i. Project Steering Committee (PSC)

ii. Independent Safety, Review and Evaluation Committee (ISREC)

iii. Project Management Team (PMT)

iv. Expert Technical Working Groups (ETWG’s)

1. Claims analysis and data collection

2. Human factors/operations/safety

3. Bus interfaces/systems/maintainability

4. Autonomous systems and controls

5. Testing and certification

v. Establish group functionality

1. Charters

2. Procedures

3. Meeting Schedules

4. Products and Reporting

5. Staffing

6. Logistical Support

b. Prepare initial technical white papers for ETWG’s

c. Schedule kickoff meeting including PSC/PMT/ISREC/ETWG’s

d. Prepare periodic invoices/time and expense sheets/financial reports as required

e. Prepare quarterly progress reports

f. Schedule monthly PMT teleconferences

g. Schedule PSC/PMT teleconferences

h. Schedule review meetings including PSC/PMT/ISREC/ETWG’s

i. Schedule ETWG meetings and teleconferences as needed

j. Prepare draft and final report

2. Describe the domain- Analysis of claims and incidents

a. Study design and statistical analysis

i. Conduct literature review

ii. Determine data availability

iii. Select sample agencies

iv. Determine criteria for models, sensitivity analysis and validation

b. Macro-level review of data available in the National Transit Database

i. Evaluate input data quality from sample of participating transit agencies

ii. Conduct comparative analysis to determine similarities and differences among

agencies to develop peer groups for predictive modeling

iii. Analyze trends in claims and identify correlations among potential causal variables

c. Micro-level agency data review and data collection

i. Link individual claims to collisions

ii. Establish framework to account for all costs attributable to collisions

iii. Collect sample data for all costs linked to individual collisions by type

-9-

d. Develop statistical models/actuarial models

i. Prepare database

ii. Quality control and data validation

iii. Model calibration

iv. Sensitivity testing and review

e. Update models with data from years 1-4

3. Development of Functional Requirements and Standards

a. Determine performance requirements for ACAT systems

i. Examination of collision types and factors influencing severity

ii. Define vehicle performance dynamic parameters

1. Allowable g-forces on passengers

2. Braking and acceleration rates

3. Steering and lateral acceleration

iii. Define Operator/human factors interactions with ACAT systems

iv. Define environmental and physical requirements for system performance

v. Define data recording and logging requirements

b. Examine available ACAT Technology

i. Examine current market products

1. Automobiles

2. Trucks

3. Buses

4. Portable applications

ii. Determine applicability of current products and evaluate effectiveness

c. Solicit vendor input from system developers

d. Prepare draft standards for bus ACAT systems

i. Performance requirements

ii. Electromechanical interfaces

iii. Data interfaces

iv. Electrical and electro-magnetic radiation

v. Environmental requirements

vi. Physical requirements

vii. Human factors/operator/maintenance requirements

e. Prepare plan for maintenance and update of standard

4. Prototyping and Testing

a. Develop safety, testing, and data logging protocols

b. Select representative vehicle types for prototyping

c. Set up test facility

d. Develop CAD models for existing vehicles

e. Develop test bed vehicle configuration design(s)

f. Develop physical test bed mock-up(s)

g. Develop ACAT functional simulator and data logger

h. Conduct static testing of ACAT simulator on test bed vehicle(s)

i. Conduct test track testing of ACAT simulator and test bed vehicle(s)

j. Conduct non-revenue over the road testing of ACAT simulator and test bed vehicle(s)

-10-

k. Approve test bed vehicle(s) for pilot vendor prototype product testing

-11-

Preliminary Draft Research Plan for Application of Autonomous Collision Avoidance Technology (ACAT) to Transit Buses schedule Version 131121 0904

Year 1 Year 2 Year 3 Year 4

Quarter 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Phase Activity

1. Create management and oversight structure

1.a Organization/Project Mobilization

1.b Prepare initial technical white papers for ETWG’s

1.c Schedule kickoff meeting PSC/PMT/ISREC X

1.d Prepare periodic invoices/financial reports as required

1.e Prepare quarterly progress reports X X X X X X X X X X X X X X

1.f Schedule monthly PMT teleconferences

1.g Schedule PSC/PMT teleconferences X X X X X X

1.h Schedule review meetings PSC/PMT/ISREC/ETWG’s X X X

1.i Schedule ETWG meetings/teleconferences as needed

1.j Prepare draft and final Report

2. Describe the domain- Analysis of claims and incidents

2.a Study design and statistical analysis

2.b Macro-level review of National Transit Database data

2.c Micro-level agency data review and data collection

2.d Develop statistical/actuarial models

2.e Update models with additional data from Years 1-4

3. Functional Requirements and Standards

3.a. Develop performance requirements for ACAT systems

3.b Examine available ACAT Technology

3.c Solicit vendor input from system developers

3.d Prepare draft standards for bus ACAT systems

3.e Maintenance plan and update of standard

4. Prototyping and Testing

4.a Develop safety, testing, and data logging protocols

4.b Select representative vehicle types for prototyping X

4.c Set up test facility

4.d Develop CAD models for existing vehicles

4.e Develop test bed vehicle configuration design(s)

4.f Develop physical test bed mock-up(s)

4.g Develop ACAT functional simulator and data logger

4.h Static testing of ACAT simulator on test bed vehicle(s)

4.i Track testing of ACAT simulator/test bed vehicle(s)

4.j Road testing ACAT simulator and test bed vehicle(s)

4.k Approve for pilot vendor prototype product testing X

PSC=Project Steering Committee PMT=Project Management Team ISREC=Independent Safety, Review & Evaluation Committee ETWG= Expert Technical Working Group

-12-

PROJECT STAFFING TABLE

Project Staffing Table Version 131121 0950

Position Title Source Year 1

Hours/%

time

Year 2

Hours/%

time

Year 3

Hours/%

time

Year 4

Hours/%

time

Principal Investigator ALK

Princeton

Visiting ?? JML

Project

Manager/Admin

Princeton

Faculty Lead –

Collision Analysis

Princeton

Faculty Lead –

Functional Control

Requirements

Princeton

Expert-Actuarial

Modelling

Consultant

Expert – Vehicle

Standards/ETWG

Coordinatior

APTA

Expert-Autonomous

Control Systems

Consultant

Expert – Human

Factors

Consultant

Expert – Transit Bus

Ops & Maintenance

Consultant

or Transit

Agency

Expert – Safety &

Testing

Consultant

Mechanical

Engineer- Buses

?

Electrical Engineer-

Automotive

?

Vehicle Technician ? Transit

Agency

Grad Student

Collision Analysis

Princeton

Undergraduate

Collision Data

Collection

Princeton

Grad Student

Prototype

Development/Testing

Princeton

Undergraduate

Prototype

Development/Testing

Princeton

Undergraduate

Prorotype

Development/Testing

Princeton

-13-

BUDGET

(to be developed)

-14-

REFERENCES

i “Understanding Transit Accidents Using the National Transit Database and the Role of Transit

Intelligent Vehicle Initiative Technology in Reducing Accidents,” Final Report June 2004, Prepared by:

Research and Special Program Administration Volpe National Transportation Systems Center Office of

Safety and Security Accident Prevention Division, DTS-73 Prepared for: Office of Research,

Demonstration, and Innovation, Federal Transit Administration and ITS Joint Program Office, Federal

Highway Administration, Washington, DC 20590, Report Numbers: FTA-TRI-11-04.1 FHWA-JPO-04-

042. Accessed June 19, 2013 http://ntl.bts.gov/lib/jpodocs/repts_te/13994_files/13994.pdf

ii T. Dunn, et al., “Assessing the Business Case for Integrated Collision Avoidance Systems on Transit

Buses,” Booz Allen Hamilton, Inc., for Federal Transit Administration, U.S. Department of

Transportation, Washing ton, D.C., 20590, August, 2007, Accessed July 31, 2013

http://www.fta.dot.gov/documents/Transit_IVBSS_Business_Case_Analysis_Final_Report_9-07.pdf

iii Noblis, Inc., “Transit Vehicle Collision Characteristics for Connected Vehicle Applications Research,”

for USDOT, Research and Innovative Technology Administration (RITA), ITS Joint Program Office

(ITS-JPO, January 2013