A PROACTIVE APPROACH TO ROAD SAFETY ANALYSIS Charles Chung (Brisk - - PowerPoint PPT Presentation

▶
a proactive approach to
SMART_READER_LITE
LIVE PREVIEW

A PROACTIVE APPROACH TO ROAD SAFETY ANALYSIS Charles Chung (Brisk - - PowerPoint PPT Presentation

Mar 30, 2024 385 likes •1.34k views

A PROACTIVE APPROACH TO ROAD SAFETY ANALYSIS Charles Chung (Brisk Synergies) Franz Loewenherz (City of Bellevue) James Barr (Miovision) LEARNING OBJECTIVES 1. How can we use traffic conflict analytics to inform proactive actions for improved

slide-1
SLIDE 1

A PROACTIVE APPROACH TO ROAD SAFETY ANALYSIS

Charles Chung (Brisk Synergies) Franz Loewenherz (City of Bellevue) James Barr (Miovision)

slide-2
SLIDE 2

LEARNING OBJECTIVES

1. How can we use traffic conflict analytics to inform proactive actions for improved road safety? 2. How can we use video analytics and machine learning systems to detect conflicts? 3. How can we work together to move towards Vision Zero?

2

slide-3
SLIDE 3

Franz Loewenherz Principal Planner City of Bellevue, WA

3

slide-4
SLIDE 4

WORLDWIDE: TRAFFIC FATALITIES

Leading Causes of Death (2004)

4

slide-5
SLIDE 5

USA: TRAFFIC FATALITIES

NHTSA, Impact of Crashes (2010): Economic Cost: $242B; Societal Harm: $836B

5

slide-6
SLIDE 6

TRADITIONAL CRASH REPORTING PROCESS

6

slide-7
SLIDE 7

CRASH BASED APPROACH: LAKEMONT INTERCHANGE CASE STUDY

In 2013, WSDOT built a new roundabout at the intersection of the WB I-90 on- and off- ramps and WLSP SE/180 Ave SE. From 2005 through 2010 there were 60 collisions recorded by the Bellevue Police Department and the WSP at this location.

7

slide-8
SLIDE 8

VISION ZERO: REFRAMING TRAFFIC DEATHS & INJURIES AS PREVENTABLE

8

slide-9
SLIDE 9

CONFLICT-BASED APPROACH: DON’T WAIT FOR CRASHES TO HAPPEN

Hyden’s Safety Pyramid (adapted from Hyden, 1987)

9

slide-10
SLIDE 10

CONFLICT-BASED APPROACH: PUBLIC INVOLVEMENT STRATEGY

Total Points Placed Ped Facilities 514 32% Bike Facilities 573 35% Ped Behaviors 57 4% Bike Behaviors 22 1% Car Behaviors 452 28% Total 1618

10

slide-11
SLIDE 11

CONFLICT-BASED APPROACH: VIDEO ANALYTICS STRATEGY

Leverage a city’s existing traffic camera system to simultaneously:

  • monitor counts and travel speed of all road

user groups (vehicle, pedestrian, and bicycle);

  • document the directional volume of all road

user groups as they move through an intersection; and,

  • assess unsafe “near-miss” trajectories and

interactions between all road user groups.

11

slide-12
SLIDE 12

PARTNERSHIP MOMENTUM

12

slide-13
SLIDE 13

PARTNERSHIP APPROACH

  • Milestone 1: Demonstrate the capability of vision technologies by detecting

relevant events in the sample traffic videos (e.g., detecting cars, pedestrians, and bikes and tracking their movements).

  • Milestone 2: Demonstrate an end-to-end system that will, continuously in

real-time, detect and store the events, and present aggregated information.

  • Milestone 3: Pilot deployment of end-to-end system (running on servers

provided by Microsoft) in the City of Bellevue traffic control center. The system will run off of a live feed.

  • Milestone 4: Support additional scenarios (e.g., near-collisions of cars with

pedestrians and bikes or patterns of bikers crossing a busy intersection).

13

slide-14
SLIDE 14

TURNING MOVEMENT COUNTS SAMPLE: 116TH NE & NE 12TH

14

slide-15
SLIDE 15

OBJECT CLASSIFICATION ACCURACY

15

slide-16
SLIDE 16

HOW NEURAL NETWORKS WORK

16

slide-17
SLIDE 17

TRAJECTORY DETECTION & TURNING MOVEMENT COUNTS

17

slide-18
SLIDE 18

VOLUME CHARTS

18

slide-19
SLIDE 19

NEAR-MISS DETECTION

19

slide-20
SLIDE 20

NEAR-MISS DETECTION

20

slide-21
SLIDE 21

JANARY 2017: COLLECT PRE-RECORDED TRAFFIC CAMERA FOOTAGE

21

slide-22
SLIDE 22

FEBRUARY-MARCH 2017: FINALIZE VIDEO ANNOTATION USER INTERFACE

22

slide-23
SLIDE 23

SPRING 2017: LAUNCH PUBLIC FACING WEBPAGE

23

slide-24
SLIDE 24

SPRING 2017: INVITE PUBLIC TO PARTICIPATE

24

slide-25
SLIDE 25

SUMMER 2017: CLASSIFY NEAR-MISS EVENTS

Time to Collision (Matsui et al., 2013) Post Encroachment Time (Van der Horst et. al., 2014) Swedish Conflict Technique (Hyden et. al., 1987)

25

slide-26
SLIDE 26

James Barr Senior Product Manager Miovision

26

slide-27
SLIDE 27

Credit: AP Photo/Seth Wenig

slide-28
SLIDE 28
slide-29
SLIDE 29

Credit: MIT Technology Review

slide-30
SLIDE 30
slide-31
SLIDE 31
slide-32
SLIDE 32
slide-33
SLIDE 33
slide-34
SLIDE 34
slide-35
SLIDE 35
slide-36
SLIDE 36

Feeding A Neural Network

slide-37
SLIDE 37

Configuration Computer Vision Verification & Correction Video in Data out

slide-38
SLIDE 38

Configuration Computer Vision Verification & Correction Video in Data out

slide-39
SLIDE 39

Configuration Computer Vision Verification & Correction Video in Data out

slide-40
SLIDE 40

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-41
SLIDE 41

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-42
SLIDE 42
slide-43
SLIDE 43

Real-world data

slide-44
SLIDE 44

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-45
SLIDE 45
slide-46
SLIDE 46

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-47
SLIDE 47

Train Evaluate

slide-48
SLIDE 48

Data Sources

Usage Collection Method Fidelity Evaluation Vehicle Counting Tools

  • Approx. Location /

Time Evaluation / Training Source CV Locates Vehicle Human Verifies and Corrects

  • Approx. Location /

Exact Time Training CV Suggests Boundaries Human Verifies and Corrects Exact Boundaries / Exact Time

Effort Quantity

slide-49
SLIDE 49

Data Sources

Usage Collection Method Fidelity Evaluation Vehicle Counting Tools

  • Approx. Location /

Time Evaluation / Training Source CV Locates Vehicle Human Verifies and Corrects

  • Approx. Location /

Exact Time Training CV Suggests Boundaries Human Verifies and Corrects Exact Boundaries / Exact Time

Effort Quantity

slide-50
SLIDE 50

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-51
SLIDE 51

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-52
SLIDE 52

Strategy

Decomposition Representation Video

slide-53
SLIDE 53

Formulation

slide-54
SLIDE 54
slide-55
SLIDE 55
slide-56
SLIDE 56

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-57
SLIDE 57

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-58
SLIDE 58
slide-59
SLIDE 59
slide-60
SLIDE 60

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-61
SLIDE 61

Configuration Computer Vision Verification & Correction Video in Data out Training

slide-62
SLIDE 62

Continuous Learning

slide-63
SLIDE 63

Continuous Learning

slide-64
SLIDE 64

Big Data

ImageNet 14,197,122 samples 132 hours of video (30 fps) Miovision 2016 Average

  • ver 16,000 hours per week

Peak Season

  • ver 10,000 hours in a

day

slide-65
SLIDE 65
slide-66
SLIDE 66
slide-67
SLIDE 67

Credit: Toronto Star

slide-68
SLIDE 68
slide-69
SLIDE 69
slide-70
SLIDE 70

Credit: MIT Technology Review

slide-71
SLIDE 71

Credit: AP Photo/Seth Wenig

slide-72
SLIDE 72

Charles Chung CEO Brisk Synergies

72

slide-73
SLIDE 73

AGENDA

  • Company introduction
  • Case studies of safety analyses
  • Types of deployment

−On-demand Safety-as-a-Service −Continuous traffic monitoring platform

73

slide-74
SLIDE 74

ABOUT BRISK SYNERGIES

  • Software firm offers solutions for improving urban mobility and

safety

  • Leader in automated traffic video safety analysis
  • HQ in Waterloo (Ontario), R&D office in Montreal
  • Clients: municipalities, DOTs and traffic consulting firms

74

slide-75
SLIDE 75

EXAMPLES OF NEAR MISSES

75

slide-76
SLIDE 76

TORONTO CASE STUDY

Road safety improvement measurements

76

slide-77
SLIDE 77

TORONTO CASE STUDY: WIDE CURB RADII

  • Location: Davenport / Christie
  • In 5 years, 2 fatal collisions &

numerous near-misses reported

  • Potential cause: high-speed

right-turn vehicles

  • Put signs with uncertain

improvements

  • Implemented curb radii

reduction with before-after safety study

77

slide-78
SLIDE 78

BEFORE AND AFTER ANALYSES

  • 6 days of before and after data collected (7am to 7pm)
  • Before data collected Aug ‘16
  • After data collected Nov ’16

78

slide-79
SLIDE 79

ANALYZED RESULTS: SPEED DISTRIBUTION

79

slide-80
SLIDE 80

ANALYZED RESULTS: CONFLICT ANALYSIS

High risk conflicts (<=1s) After: 9 instances Before: 19 instances

2 4 6 8 10 12 14 16 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

PET Before/After Conflicts

Before After

80

slide-81
SLIDE 81

QUANTIFYING IMPROVEMENTS

  • Pre-normalized results
  • High Risk Conflict Rate are reduced by 30%
  • Medium Risk Conflict Rate are reduced by 38%
  • Low Risk Conflict Rate are reduced by 72%

Low Risk Conflict Medium Risk Conflict High Risk Conflict Count Rate Count Rate Count Rate Before 58 93,843 24 38,831 19 30,742 After 11 26,465 10 24,059 9 21,653

81

slide-82
SLIDE 82

TRAJECTORY HEATMAPS

Before After Cars Pedestrians

82

slide-83
SLIDE 83

OTHER SCENARIOS

83

slide-84
SLIDE 84

NYC: FAIL-TO-YIELD DETECTION

84

slide-85
SLIDE 85

HIGHWAY CONFLICT ANALYSIS

85

slide-86
SLIDE 86

CONGESTION ANALYSIS CASE STUDY

86

slide-87
SLIDE 87

ILLEGAL RIGHT-TURNS

87

slide-88
SLIDE 88

JAYWALKING BEHAVIOURS

88

slide-89
SLIDE 89

ANALYZED RESULT

Conflict hotspots 20-sec Conflict Videos Heatmaps

5 10 15 20 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Count PET (seconds)

CONFLICT DISTRIBUTION

89

slide-90
SLIDE 90

ON-DEMAND ANALYSIS

On-demand Analysis Service

  • Data collection by Brisk
  • Speed, count, conflict, etc.
  • Result and report in 2 weeks

Traffic Monitoring Platform

  • Access connected cameras

(TMC)

  • Continuous analysis
  • Historic results on web

90

slide-91
SLIDE 91

ON-DEMAND ANALYSIS

On-demand Analysis Service

  • Data collection by Brisk
  • Speed, Count, Conflict, etc
  • Result and Report in 2

weeks Traffic Monitoring Platform

  • Access connected cameras (TMC)
  • Continuous analysis for years
  • Historic results on dashboard

91

slide-92
SLIDE 92

CONTINUOUS MONITORING CASE STUDIES

  • City of Montreal
  • 5 TMC connected-camera
  • 20 scenarios of movement interactions
  • Conflict and non-conflict scenarios
  • Veh/Veh, Veh/Peds and Veh/Cyclists

92

slide-93
SLIDE 93

CONTINUOUS MONITORING CASE STUDIES

  • Roundabout at Region of Waterloo
  • Frequent collisions
  • Monitors right-of-way violation
  • Track improvements of treatments

93

slide-94
SLIDE 94

94

Franz Loewenherz

Principal Planner

FLoewenherz@Bellevuewa.gov

James Barr

  • Sr. Product Manager

JBarr@miovision.com

Charles Chung

CEO

Charles.Chung@brisksynergies.com