Intersection Decision Support IDS: Infrastructure and Cooperative Approaches to Intersection Safety

1
Intersection Decision Support (IDS)
Infrastructure and Cooperative Approaches to
Intersection Safety

• James A. Misener
• PATH Program-Wide Research Meeting
• October 25, 2002

2
Outline

• Introduction
• The IDS Team
• Problem Definition
• National Approach
• Program Plan
• Discussion of Spring, 2003 IDS Demo
• Research Topics
• Challenges
• California Portion (by example)
• Initial Scenarios
• Systems Architecture
• Intersection Testing
• Driver Infrastructure Interface
• Wireless Communications
• June 2003 Demo
• Summary
• Final Notes

3
The IDS Team IVI Infrastructure Consortium

• IVI Infrastructure Consortium Principals
• California DOT (Caltrans)
• Minnesota DOT
• Virginia DOT
• USDOT (FHWA)
• Universities conducting the IDS research
• U.C. Berkeley (California PATH, other units)
• University of Minnesota (ITS Institute, other
  units)
• Virginia Tech (VTTI)

4
The IDS Team (Contd)

• Initially was IVI Specialty Vehicle Platform
  pooled-fund project
• USDOT requested transformation into IVI
  Infrastructure Consortium in 1999
• Parallel to vehicle industry IVI Enabling
  Technologies Consortium

5
IDS Management Structure
Infrastructure Consortium California DOT
(Lead) Minnesota DOT Virginia DOT
IDS Leadership Group (Co-PIs) Dr. Steven
Shladover (Chair) Prof. Max Donath Dr. Vicki
Neale
IDS Project Manager Mr. James Misener
U.C. Berkeley Site Manager Mr. James Misener
Virginia Tech Site Manager Dr. Vicki Neale
U. of Minnesota Site Manager Dr. Craig
Shankwitz
6
Problem Definition

• 27.3 of all police-reported crashes ?
• 1.72 M crashes
• 9000 fatalities/year ? Almost 25 of all
  traffic fatalities
• 1.5 M injuries/year ? Approximately 50 of all
  traffic injuries

• Problem Unacceptably high frequency and
  severity of crashes at intersectionsby almost
  any measure

7
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Crash Scenario
• Straight crossing path (SCP) 36.6

8
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Crash Scenario
• Left turn across path/opposite dir.
  (LTAP/OD) 27.3

9
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Crash Scenario
• Left turn across path/lateral dir. (LTAP/LD) 15.9

10
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Crash Scenario
• Left turn in path (LTIP) 4.7

11
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Crash Scenario
• Right turn in path (RTIP) 4.7

12
Intersection Crash Statistics (Smith and Najm,
based on 1998 GES)

• By Traffic Control Device
• 3-phase signals 41.6
• Stop signs 36.3
• No controls and other 22.1

13
But Would an IDS Work?

• Distance-wise, yes (Ferlis, 2002)
• Timing-wise, yes (Lerner, 1995 Ferlis, 2002)
• Technology-wise and institutionally? This is
  reason for US DOT and three States to join forces

14
A Snapshot of the IDS Approach

• Emphasis on
• Exploiting Cooperative Systems (infrastructure
  in-vehicle)
• Key Question(s)
• Where do the sensing/processing/decision-to-warn/w
  arning occur?
• Infrastructure or vehicle?
• Systems Approach
• Produce Nationally Interoperable Architecture
• Driver Decision Aids (e.g., gap, velocity
  information)
• Crossing-Path Collisions (Signalized and
  Unsignalized)
• 78.1 of Intersection Crashes (1998 GES)

15
Approach (Contd)

• Focus on Driver Recognition and Driver
  Decision error crash causes
• Hypothesis Provide the driver with information
  that will improve judgment of gap clearance and
  timing
• Optimize information and HMI
• Examples gap and TTC feedback
• Secondary concerns Driver Erratic Action and
  Drunk Driver

16
IDS Program PlanItalics Denote Currently Active
Tasks

• 0 Detailed Program Planning
• A Delineate the Intersection Crash Problem
• B Develop Top Level Requirements for
  Types/Classes of Intersection Crashes
• C Conduct Enabling Research Development
• Under this task we will conduct demonstration at
  Turner Fairbank Highway Research Center
• - June, 2003
• Emphasis Infrastructure- and cooperative
  (comm.)-based IDS

17
IDS Program Plan (Contd)

• D Prioritize Classes of Intersection Crashes
  for Initial Study
• E Conduct Countermeasure Trade-off Analyses
• F Develop Detailed Requirements and
  Specifications for Each Countermeasure/Crash
  Class
• G System Design and Development
• H Conduct Subsystem Tests and Experiments
• I Prepare for Countermeasure Demonstration

18
Research Topic Coverage
19
Initial Scenarios
20
California Approach

• Original (and Preferred) IDS Concept
• Systems engineering and analytical process to
  identify highest value problem(s) and solution(s)
• Nationally interoperable architecture and
  components
• Current Methodology
• Look at Symptoms Examine distribution of
  scenarios by various methods
• Number of Crashes
• Fatalities
• Injuries
• Cost using recent NHTSA analytical techniques
• Determine Common Causes Examine distribution
  of fundamental driver causes
• From 1 and 2, Determine Engineering Solution(s)
  for Groups of Causes
• Determine Preferred Scenario, Which
• Addresses most symptoms
• Clusters causes
• Is amenable to ITS solutions

21
California Scenarios LTAP/OD and LTAP/LD
(Urban)
22
Systems Architecture
23
Vehicles
DSRC
Infrastructure
Traffic Signal
Traffic Signal controller
messages
DII manager
Traffic Signal manager
Conflict predictor
Gap predictor
Stop predictor
Future State
Future State Predictor
State map
State map
State Map Generator
State Map Generator
The dynamic intersection map is broadcasted to
the vehicles / the vehicle MAY send information
to the infrastructure
position, speed, etc
position, speed, signal state, time to next
phase, etc
Vehicles sensors
Infrastructure sensors
Demo Physical Architecture
24
(No Transcript)
25
Physical Architecture Smart cars / nonequipped
intersection
26
Intersection Testing
27
(No Transcript)
28
Intersection Features

• Funded and Designed to Caltrans Standards Under
  Separate, Complementary Funding
• Standard Features
• Four-legged Intersection
• Roadway extension
• Right turn lane and stripped island for
  eastbound approach
• Each leg with two 3.6-meter lanes
• One lane for each direction
• 1.5-meter shoulders
• 3.1-meter pedestrian crosswalks
• Standard Caltrans roadway striping
• Two Street light poles on opposite sides of
  intersection
• Signs Signals
• Coverable STOP Signs on signal poles
• Removable STOP Signs to be placed on pavement
• Standard Caltrans signal heads
• ITS Control cabinet (340) with 170 2070
  controllers
• Handicapped accessible pedestrian push buttons
  and signs

29
Intersection Features (Contd)

• Enhanced IDS Features
• Loop Detectors
• 3M Microloops
• Moveable Nu-Metrics (wireless) loops
• Standard embedded loops
• Combination of the above loop detectors
• Driver Infrastructure Interface (DII)
• Placed on boundary of one approach leg
• Six possible locations for DII, five on
  northbound approach on Wren drive and one on the
  opposite side of intersection
• Opaque Fence
• Placed on boundary of one approach leg
• 1.3-meter in height and 60-meter in width
• Fabric over a frame
• Visual Background
• Depending on approach, tree-filled or with some
  buildings

30
Intersection Features (Contd)

• Enhanced IDS Features (Contd)
• Wireless Antennas
• One directional antenna
• One omni-directional antenna
• Cabling
• Access point cabinet

31
Naturalistic Data CollectionGoals and Research
Questions

• Goals
• Collect information about drivers behavior while
  approaching and crossing an intersection
• Research questions aimed at decision making when
  approaching and crossing intersections
• Visual search?
• Gap selection and execution?
• Information needs?

32
Instrumented Vehicle
33
Preliminary Protocol

• Test of participant vision and questionnaires on
  driving practice 1 hour
• Driving session on an itinerary including a set
  of representative intersections (selected based
  on Task A inputs) 1 hour
• Interview post-driving sessions on a selected set
  of intersections 1 hour

34
Data processing
synchronized video and engineering files
Radar file time, target id
165425.658
Sensor file time, throttle
Situation file time, light cycle,
35
Implementation

• Intersection geometry and motion models derived
  from TEXAS
• Developed by T. Rioux at UTexas
• ftp//ftp.ce.utexas.edu/texas_model
• Some vehicle and environmental models from
  SmartAHS
• http//www.path.berkeley.edu/smart-ahs
• Additional models from simulation subtasks
• Driver
• Communication
• Infrastructure
• Sensors

36
Driver Infrastructure Interface
37
Wireless Communication
38
Example IDS with Wireless Communication

• X is equipped with a radar AND with a radio and a
  GPS and it is approaching an intersection

4
1
x
2
3
39
An example contd
4
1
x
2
3
X state Map
1
2
me
40
Communication used to reduce sensor errors
4
1
x
2
3
X state map
1
2
me
41
Communication used to overcome sensor limitations
4
1
x
2
3
X state map
4
1
2 sec to red
2
3
me
42
At the same time communication my be used to
deliver WARNING messages
4
1
x
2
3
X state map
4
1
2 sec to red
2
3
me
43
June 2003 Demo
44
LTAP/OD Demo
45
Summary

• Intersection Decision Support should improve
  safety at intersections
• Will facilitate better driver decisions
• Program plan is broad and well-developed
• Systems approach, based on problem definition
• Balances Infrastructure and Cooperative
  approaches
• Diverse national problem needs nationally
  interoperable solution
• Three-year program will deliver results
  continuously
• Identify near-term deployment opportunities
• Culminates in FOT plan(s)

46
Final Notes

• California Prefers More Considered, Systematic
  Approach
• Our Approach Now Emphasizes Certain Crash
  Typologies
• LTAP/OD, LTAP/LD
• Some consideration for dilemma zone and
  cross-cutting cooperative applications
• Current Progress in
• Epidemiological studies
• System architecture
• Communications
• Intersection at PATH
• Other areas
• We Refocus Certain Areas
• Our own remote surveillance investigation
• Add cost/benefit analyses for CA scenarios
• Reduced level of effort for systems engineering

47
Questions? Or a Movie?