“Towards a Multi-state Consensus on Rural Intersection Decision Support”

Pooled Fund MeetingMinong, Wisconsin

June 12, 2006

From IDS to CICAS:Rural Intersection Crash Avoidance

National Motivation

2.6 million intersection related crashes annually Represents 41% of all 6.33 million police reported

crashes In Minnesota, 129 out of 583 (22.1%) fatal crashes are at

intersections In the US, 8,659 of 38,252 (22.6%) of fatal crashes were

intersection related 31.7% occurred at signalized intersections 68.3% occurred at unsignalized intersections (stop

sign, no controls, other sign).

NHTSA, Traffic Safety Facts 2003, January 2005Minnesota Office of Traffic Safety, Minnesota Motor Vehicle Crash Facts, 2003

Intersection Decision Support (IDS)

Focus on driver error causal factors Fatal and life changing intersection crashes Provide the driver with information that will improve

judgment of gap clearance and timing Keep major corridors flowing Deploy where the fatalities/crashes warrant

deployment New tool for the traffic engineer

Focused on Recognized National Problem

NCHRP Report 500:Vol. 5 Unsignalized Intersections

Identifies objectives and strategies for dealing with unsignalized intersections

Objective 17.1.4 Assist drivers in judging gap sizes at Unsignalized Intersections

High speed at grade intersections

Guidelines for Implementation ofAASHTO Strategic Highway Safety Plan

Approach

Measure gaps that drivers take under actual road conditions. Collect data regarding intersection entry behavior.

Evaluate suite of sensors to ensure that they are able to measure those gaps accurately and able to distinguish between unsafe and safe gaps at the level that the existing literature specifies.

Develop set of interface concepts with which to communicate the existence of an unsafe vs safe gap in the intersection to the driver.

Evaluate selected set of interface concepts.

Percentage of all crossing path crashes based on Najm W G, Koopmann J A, and Smith D L (2001) Analysis of Crossing Path Crash Countermeasure Systems. Proc. 17th Intl Conference on Enhanced Safety of Vehicles

TCD Factor LTAP/OD LTAP/LD LTIP RTIP SCPGap 16.11% 1.09%Violate 2.59% 4.34% 1.25% 0.50% 14.86%Gap 1.25% 9.43% 2.17% 2.09% 14.44%Violate 0.08% 9.35% 0.58% 0.25% 5.18%

None Gap 7.68% 2.09% 0.83% 0.92% 2.92%Stop Sign

Traffic Signal

Unsafe gap is main risk

factor

Goal

Design and evaluate information “concepts” for sign elements to support detection and acceptance of safe gaps in mainline traffic.

Explore out of the “toolbox” concepts.

Prohibitive frame: Provide information regarding unsafe gaps.

Final judgment of safety and responsibility for action must remain with driver.

Concepts

BaselineDriver recognize hazard, gather information, decide on safety condition, and choose action

AlertDriver must gather information, decide on safety condition, and choose action.

DisplayDriver must decide on safety condition, and choose action.

WarnDriver must choose action.

AdviseDriver must choose to comply.

System detects hazard.

System detects hazard & presents information relevant to vehicle gap. Prohibited actions also indicated.

System detects hazard and provides warning levels based on gap information. Prohibited actions also indicated.

Prohibited actions indicated (unsafe action advisory).

InformDetect

Safe Gap Definition

12.5 s8.0 s

7.5 s

AASHTO Green Book, 2001; FHWA Older Driver Handbook, 2001

7.5 s

2-stage crossing strategy- Driver stops in median

1-stage crossing strategy- Driver does not stop in median

System accounts for worst-case scenario of an older driver making a left turn (in 1-stage).

Assumptions

Many factors determine gap safety. As a non-cooperative system, these sign concepts do not

have “preview” of all these factors. Therefore, system is “blind” and must make assumptions

about condition. With vehicle classification, will be able to adjust gap for

vehicle. For liability reasons, sign concepts must assume worst-

case condition. Left turns + older drivers + 1-stage

This may be perceived as non-credible to drivers in all other conditions.

Virtual Environment for Surface Transportation Research

•8 channels •3D surround sound•Car body vibration•Force feedback steering•Power-assist feel on the brakes•3-axis electric motion system

• Ability to model precise reproductions of geo-specific locations

• Resolution = 2.5 arc-minutes per pixel

Methods

5 sign conditions Within-subjects

2 age groups 24 young; 24 old

Gender balanced 2 light conditions

Day vs. night

12 participants per group/condition

Main Lessons

Additional information can be used. All sign concepts resulted in shifts towards the safe gap

threshold However, threshold was perceived as too conservative (not

personalized) Compliance increases when visibility of traffic condition

is reduced. Old drivers, night condition

Dynamic aspects of (Icon and Split-hybrid) signs facilitate comprehension. Map sign changes to changes in environment

Young drivers may “calibrate” system.

Next Steps

“Personalized” gap thresholds Cooperative systems

MUTCD compliant formats Close interaction with MnDOT engineers

Test compliant formats in simulator Validate with test site experiment Field operational testing Consider role of signs in larger safety programs

(Training & Education). Older drivers must perceive own limitations to appreciate need

for decision support Drivers need understanding of functions (include in licensing

tests)

Pooled Fund Study(CA, GA, IA, MI, MN, NC, NH, NV, WI)

Goals: Characterize rural intersection crashes throughout

USA Identify regional differences in driver behavior Use information to design ubiquitous system

deployable throughout US Setup a broad base for a field operational test

Key Tasks: Crash analysis in each partner state Driver behavior data collection in each partner state Analyze and archive driver behavior data

Cooperative Intersection Collision Avoidance System (CICAS):

Stop Sign Assist

CICAS

CICAS Work Plan (5 Years)

IDS CICAS

GapModel

SituationAnalysis

Concept Study

TranslationStudy

Sign Concepts

Paramete

rs

Functional Scope

ValidationStudy Pre-FOT

FOT

Protocol S

tandardization

Protocol E

valuation

Compliant Signs Deployable Signs

Alert / TimingAlgorithms

CAMPDVI

Years One thru Three Years Four and Five

Situation Analysis

Macroscopic Common scenarios Outliers

Onsite observation Context Atypical cases

Crash reports Common risk factors

Gap Model / Algorithms

3.8 3.94.3

4.7

0.0

1.0

2.0

3.0

4.0

5.0

0 .01 - .25 .25 - .9 .9 - 1.5

Precipitation (cm/hr)

Gap

(s)

- 5

th P

erce

nti

le

Macroscopic License plate reader Demographics

Microscopic Instrumented vehicle Process stages

Predictive models Sensitivity analysis Practical analysis

Algorithms Gap threshold, timing Complete logic

• Platoons, non-cooperative cases, etc.

Translation Study

Integrate algorithm Convert “concepts” to

compliant signs• Inform, warn, and advise

Test legibility and comprehension.

Replicate sim evaluation• Do compliant signs retain

concept benefits?

Interaction with DVI?

Technical Steps to FOT/Deployment

Minimal Sensor Sets Mainline: Function of variation in mainline traffic speed

and sensitivity of driver to timing (HF phase)• Have data showing speed variation and comparison

to point sensors. Minor road: Function of sensitivity of vehicle type to

gap alert/warning timing• Definition of gap previously used shows little

sensitivity to vehicle type. • Microscopic study of driver behavior needed to

resolve this issue.

Technical Steps to FOT/Deployment

Driver behavior data Macroscopic data

• Pooled fund: data collection in eight states with portable intersection surveillance system

Validation Study: Microscopic Data• Fully instrumented passenger car• Fully instrumented heavy truck• Fully instrumented intersections

Testing in Minnesota

Technical Steps to FOT/Deployment

Driver-Infrastructure Interface Mechanical & electrical design Placement Cost Reliability

Communication Mechanism Dedicated short range

Communications (DSRC) Wireless Access for Vehicular

Environments (WAVE; 802.11p) FCC allocated Oct. 21, 1999 75 MHz of spectrum at 5.9 GHz 7 licensed channels Hardware and protocols under

development

Technical Steps to FOT/Deployment

Definition/Implementation of ‘Cooperation’ Driver-Infrastructure Cooperative: demographic,

personal preference data• Storage (on person, in-vehicle?)• Broadcast (from person, through the vehicle?)• Auto manufacturers seem opposed to personal data

broadcast and used by infrastructure system Vehicle-Infrastructure Cooperative: vehicle

performance, weight, size, etc.• Driver intent (turn signal, steering wheel position,

foot on clutch, brake, throttle, gear selection, etc.)

Validation Study:Support of On-site Human Factors Testing

Build Instrumented Vehicle Wave-DSRC radios Integrated eye tracker Steering, brake, throttle measurements Full traffic data Day, night testing

Support analysis Location, speed, etc. of all other vehicles in

vicinity of intersection

Validation Study:On-site Human Factors Testing

Intersection DAQ (iDAQ)

Traffic Data(position, speeds, lane of

travel, etc.)

Vehicle Data (Throttle, brake, steering,

turn signal, position, speed, accel., etc. )

Driver Data (Eye Gaze, hand, feet,

face cameras, etc. )

Environmental Data (Weather (R/WIS) data,

Sun location (glare), road conditions)

Vehicle DAQ(vehDAQ)

NTP (synch)

FOT

IDS CICAS

GapModel

SituationAnalysis

Concept Study

TranslationStudy

Sign Concepts

Paramete

rs

Functional Scope

ValidationStudy Pre-FOT

FOT

Protocol S

tandardization

Protocol E

valuation

Compliant Signs Deployable Signs

Alert / TimingAlgorithms

Pre-FOT Years One thru Three FOT Years Four and Five

Goal

Provide real world data of system effect on gap acceptance and driver perceptions to support policy decisions for deployment trials.

Method

Recruit local residents using own vehicles N = 30

Male and female Young and old

Compare behavior before and after installation of system.

Plan

Methodology Data Harmonization

Pilot FOT Evaluate methodology Final design iteration

Full FOT Naturalistic scenario Macroscopic

• Microscopic?

Deployment verdict

Vehicle Computer

L. Turn Signal

Brake Light

R. Turn Signal

Trailer Interface

Kit

DSRC Radio

Driver Demo

Information

Vehicle Information

(size, type, ID)

Driver Intent

Estimator

IntersectionController

(iDAQ)

DSRC Radio

DII

Vehicle

FOT Instrumentation:Vehicle Cooperative System

12 240 36 48 60Month

Task 1: Project Management and CoordinationTask 2: Research

2A: Driver Behavior Research

2B: Driver Interface Research

2C: Models, Algorithms

Task 3: System Design3A: DII Design

3B: Minimal Sensor Sets

3C: Mechanisms for Cooperation

3D: Design Documentation

Task 4: System Development and Prototype Testing

4A: Dev. Plan

4B: Integration

4C: Objective Testing

4D: iDAQ

4E: FOT Plan & Design

Task 5: FOT Plan and Design

5B: Final FOT Design

Task 6: Conduct FOT6A: Pilot

FOT6B: Conduct FOT

Task 7: Outreach

Critical point 1DII MUTCD

approval

Critical point 2DSRC: $, license,

performance

Critical point 3Instrumentation

(gaze vector precision, accuracy)

Critical point 4Government GO/NO GO decision

Critical point 5 Recruitment of Pilot and

FOT Subjects

Critical point 6 Liability (risk management,

IRB, subpoenas)

CICAS-GAP Timeline and Critical PathT

asks

5A: FOT Vehicle & Intersection Tests