AEB Presentation - Thatcham Motor Insurance Repair Research ...

AEB
TTest t Development
D l t
Autonomous Emergency Braking

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
2

Agenda g
� AEB TTest t GGroup
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
3

AEB Test Group
Rationale for Group
� AEB systems are becoming more widely available
� Pedestrian protection is a significant issue
�� Technology is available to address pedestrian casualties
� Definition of standardised test procedures to encourage
and d control t l such h systems t
– to cover all aspects of societal impact: repair and material
damage costs, costs injury and fatality costs
4

AEB Test Group
Terms of reference
� IInternational t ti l group
� Insurer and consumer perspective
� Two year initial project (2009-2011)
– Interrogate detailed accident data
– Id Identify tif pre-crash h situations it ti relevant l t ffor current t AEB technology t h l
� Focus on injury reduction – including whiplash and
pedestrian collisions
� Also covering
– Material damage cost reduction
– Appropriate positioning and
mounting requirements for sensors
to prevent increase in insurer costs
following minor crash damage
5

AEB Test Group
Scope and aims
� “To design and implement test procedures reflecting real
world data that can encourage the development of
autonomous braking technology that can help prevent or
mitigate the effects of car-to-pedestrian and car-to-car
crashes”
CCar-to-Pedestrian t P d t i CCar-to-Car t C RRear CCar-to-Car t C HHead-On d O
(CP)
(CCR)
(CHO)
6

AEB Test Group
Phase 1 Project Partners
� Supported by:
– Loughborough
University
(Vehicle Safety
Research Centre)
– 1 x OEM
– 1 x Tier 1 supplier
7

AEB Test Group
Development and Implementation of Procedures
� Incorporate provisional results from real world accident
data to define test conditions
� Define and specify test measurement equipment
� Define test metrics
� Define rating process
� Author test procedures
� Publish initial results/ratings to inform
consumers/stakeholders of technology capability
� Integrate into existing Consumer Test Programs (RCAR)
� Off Offer to Euro NCAP C P-NCAP C ffor consideration for f future f
test program
8

AEB Test Group
Milestones & Publications
� Mil Milestones: t
– Publish initial test concept early 2011
– Feedback from industry
– Validate with physical testing 2011
– Finalise test procedures and equipment end 2011
– Aim of undertaking full tests during 2012
� Publications:
– Test procedures
p
– Current status updates at technical conferences and industry
meetings
– SSupporting ti materials t i l will ill iinclude l d presentations, t ti ttest t
documentation, test videos and others
– www.thatcham.org/AEB
t atc a o g/
9

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP): (CP)
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
10

Accidentology Study
Introduction
� DData t analysis l i and d report t authoring th i ffrom LLoughborough hb h UUniversity i it
using GB accident data:
– OTS
– STATS 19
� Data analysis from the German Insurers Accident Research (UDV)
and Insurance Institute for Highway Safety (IIHS)
� Review of other worldwide data and accidentology studies
– Japan, p , Australia, , China
� Examples of factors under investigation:
– Frequency of collisions
– Road layouts
– Speeds
– Typical pre-accident pre accident causation factors
– Sight lines
– Mitigating factors (parked cars, complex intersections etc.)
11

Accidentology gy
Introduction: UK data study
� Aim to define common accident scenarios
– pedestrian accidents
– rear-end impacts
– head-on impacts
� Source materials
– STATS 19 national accident database
– On-the-Spot in-depth accident database
STATS 19
2008
Great Britain
National statistics
Police
Police reports
All casualty accidents
170,000+ accidents
On-the-Spot (OTS)
2000–2009
South Notts and Thames Valley
2010 casualty reduction targets
Research teams
At-scene investigations
Police Police-attended attended accidents (8 hours daily)
500 accidents per year
12

Accidentology gy
Introduction
� Overlap of cases in STATS 19 and OTS
– STATS 19 wider region for shorter period (2008)
– OTS narrower region for wider period (2000–09)
– Limited overlap of cases
Region
South Notts
Thames Valley
PPeriod i d
STATS 19
OTS
13

Accidentology gy
Introduction
� Flow of work
STATS 19
OTS
Source
databases
Computer
logic
stats19
Cluster
analysis
Detailed
Cluster
case
ots analysis
reviews
Preparation
Summary
datasets
Clusters
Scenarios
14

Accidentology gy
Methodology
� Cluster analysis
– mathematical grouping of data points
1
2
3
4 5
2-dimensional space
Dendrogram
15

Accidentology gy
Methodology
� Distance between accident records (dissimilarity)
– numerical values
– nominal, ordinal and scalar types
nominal
nominal
ordinal
ordinal
scalar
scalar
Record eco d 1 Record eco d 2 Distance sta ce
1
3
0.0
1.00
0.2
1.0
Daylight
Turning*
10–30mph 10 30mph
Adult male
50km/h
11m
1
1
1.0
0.67
0.8
0.0
Daylight
Going ahead
60–70mph 60 70mph
Adult female
80km/h
3m
Pedestrian age-sex age sex (ordinal)
Vehicle speed km/h (scale)
0–7 y 8–15 y Female Male 40 50 80 90
0 0.33 0.67 1.0 0 0.2 0.8 1.0
0.00
1.00
1.00
0.33
0.60
1.00
3.93
16

Accidentology gy
Methodology
� Distance between clusters
– ‘Average of distances’ used
Closest points Farthest points points Average of distances
17

Accidentology gy
STATS 19 — Pedestrian clusters
Aim for 4-6
clusters
≥75% of cases
Example
Cluster 1
• 39% of cases
• Lower speeds
• Daylight
• Fine weather
• Going ahead
• Children
Green cells:
over-represented
• Crossing from left
99 5% in χ • Not masked
2 99.5% in χ test
• Not masked
18

Accidentology gy
OTS — Pedestrian location (1 sec) N=175
Lateral disstance
(m)
5.0
4.0
3.0
2.0
1.0
-1.0
Visible
Not visible
00 0.0
0 5 10 15 20 25 30
-2.0
-3.0
-4.0
-5.0
Forward distance (m)
19

Accidentology gy
OTS — Pedestrian clusters
20

Accidentology gy
Pedestrian scenario 1
21

Accidentology gy
Pedestrian scenario 2
22

Accidentology gy
� Direct link between accident data and clusters
– mathematical computation
– objective and reproducible
– representativeness clearly defined
� Remarkable parallels in STATS 19 and OTS
– pedestrian baseline and obscured scenarios
– rear-end stationary and moving scenarios
– estimate of representativeness
� Innovative approach and fresh results
23

Simulation Example: p Collision
Car-to-Pedestrian Case 1
� Ub Urban road d
environment:
– Residential area area,
daylight, fine weather
� VW Golf:
– Travel speed 50km/h
– Impact speed 50km/h
� Pedestrian:
– Walking; no sight
obstruction
– 22 year old male
– Height 1.8m, weight
80 kg
– Serious injury
24

Simulation Example: p AEB
Car-to-Pedestrian Case 1
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds (this is the
time before the actual
impact)
� VW Golf:
– Travel speed 50km/h
– Braking 0.8g
� Pedestrian:
– Walking 5km/h
� Vehicle stops 0.3m
bbefore f impact i
25

Simulation Example: p Collision
Car-to-Pedestrian Case 2
� Ub Urban road d
environment:
– Residential area area,
daylight, fine weather
� Ford Focus:
– Travel speed 50km/h
– Impact speed 50km/h
� Pedestrian:
– Walking—parked cars
on side of road
– 12 year old male
– Height 1.67m, weight,
61 kg
– Serious injury
26

Simulation Example: p AEB
Car-to-Pedestrian Case 2
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds (this is the
time before the actual
impact)
� Ford Focus:
– Travel speed 50km/h
– Braking 0.8g
� Pedestrian:
– Walking 5km/h
� Vehicle stops 0.3m
bbefore f impact i
27

Simulation Example: p Collision
Car-to-Pedestrian Case 3
� Ub Urban road d
environment:
– Residential area area,
daylight, fine weather
� VW Golf:
– Travel speed 50km/h
– Impact speed 35km/h
� Pedestrian:
– Running on pavement;
no sight obstructions
– Crosses road
diagonally
– 21 year old male
– Height 1.9m, weight
80 kg
– Injured
28

Simulation Example: p AEB
Car-to-Pedestrian Case 3
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds (this is the
time before the actual
impact)
� MINI:
– Travel speed 50km/h
– Braking 0.8g
� Pedestrian:
– Running 12km/h
� Vehicle stops 0.2m
bbefore f impact i
29

Simulation Example: p Collision
Car-to-Pedestrian Case 4
� Ub Urban road d
environment:
– Residential area area, dark dark,
raining
� Peugeot 306:
– Travel speed 50km/h
– Impact speed 23km/h
� Pedestrian:
– Walking—obscured by
moving vehicle
– 28 year old female
– Height 1.6m, weight
64 kg
– Serious injury
30

Simulation Example: p AEB
Car-to-Pedestrian Case 4
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.0
seconds (this is the
time before the actual
impact)
� Peugeot 306:
– Travel speed 50km/h
– Braking 0.8g
� Pedestrian:
– Walking 6km/h
� Vehicle stops 0.6m
bbefore f impact i
31

AEB Test Concepts p
Car to Pedestrian (CP)
� Pedestrian walks from
nearside pavement into
path of car travelling along
road
� Pedestrian walks from
nearside id pavement t ffrom
behind an obstruction into
path of car travelling along
road
Provisional/DRAFT only
32

AEB Test Concepts p
Car to Pedestrian (CP)
� PPedestrian d t i runs ffrom ffar
side pavement into path of
car travelling g along g road
� Pedestrian walking along
near side of carriageway in
path of car travelling along
road
Provisional/DRAFT only
33

AEB Test Concepts p
Car to Pedestrian (CP)
� Pedestrian walks across
junction from near side
pavement into path of car
turning towards far side
into junction
Provisional/DRAFT only
34

AEB Test Example p
Car to Pedestrian (CP)
� WWalking lki pedestrian d t i approaching hi ffrom near side id
� Range of test speeds from 10 to 60km/h
� SSpeed d iincreased d iin 10k 10km/h /h increments i t if
system avoids collision with pedestrian
� Speed increased in 5km/h increments if
collision occurs
� Quarter car width and half car width impact p
positions between test car and pedestrian with
no AEB
� GGoal l – obtain bt i AEB system t performance f curve
at multiple impact positions over a range of
speeds to quantify system performance
� Then perform additional scenarios at a reduced
number of speeds to investigate system
robustness
Provisional/DRAFT only
35

Test Target g and Propulsion p
Car-to-Pedestrian (CP) initial development
36

Test Target g and Propulsion p
Car-to-Pedestrian (CP)
� Adult 50th percentile male
� ~12 year y old child
� 3 dimensional
� Easily recognisable as a walking human
�� Human proportions
� Travel speed up to 15km/h
� Appropriate pp p whole body y radar cross section
� Appropriate thermal signature
� Minimal foot to ground separation
�� Typical clothing – shirt and trousers
� Repeatedly impactable at up to 50km/h
� Considerations for vision systems y
– Colour of clothing and contrast with background
– Motion of leg and/or arm
� Autonomously controlled rather than suspended from gantry
37

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
38

Accidentology gy
STATS 19 — Rear-end clusters
Aim for 4-6
clusters
≥75% of cases
Example
Cluster 1
• 30% of cases
• Lower speeds
• At junction
•Daylight yg
• Fine weather
• Veh A going ahead
Green cells:
over-represented
• Veh B stop/starting
99 5% in χ • Following traffic
2 99.5% in χ test
• Following traffic
39

Accidentology gy
OTS — Rear-end vehicle speeds N=50
40

Accidentology gy
OTS — Rear-end clusters
41

Accidentology gy
Car-to-Car Rear Scenario 1
42

Accidentology gy
Car-to-Car Rear Scenario 2
43

Simulation Example: p Collision
Car-to-Car Rear Case 1
� RRural l road d
environment:
– Crossroad junction, junction
daylight, fine
� Volvo S70 (striking):
– Too close behind;
driver does not react
in time
– Travel speed 30km/h
– Impact speed 26km/h
� Audi A4 (struck):
– Slowing on approach
tto junction j ti
– Travel speed 30km/h
– Impact speed 20km/h
44

Simulation Example: p AEB
Car-to-Car Rear Case 1
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.0
seconds (this is the
time before the actual
impact)
� Volvo S70 (striking):
– Travel speed 50km/h
– Braking 0.8g
– Impact speed 20km/h
� Audi A4 (struck):
– Travel speed 50km/h
– Braking 0.35g
– Speed at impact
3km/h
� Mitigation of impact
severity 45

Simulation Example: p Collision
Car-to-Car Rear Case 2
� RRoad d environment: i t
– T-junction, daylight,
fine
� VW Passat (striking):
– Approaching g row of
cars at junction
– Travel speed 80km/h
– IImpact t speed d 55km/h 55k /h
� Toyota Yaris (struck):
– Travel speed 50km/h
– Impact speed 11km/h
� Ford Focus & BMW 1
Series (secondary):
– Travel speed 50km/h
– Impact speed 0km/h
46

Simulation Example: p AEB
Car-to-Car Rear Case 2
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds at 0.4g
– Activation at time to
collision lli i (TTC) 0.6 0 6
seconds at 0.8g
� VW Passat (striking):
– Travel speed 50km/h
� Other vehicles
unchanged:
– Travel speed 50km/h
– BBraking ki 00.35g 35
� VW Passat stops 0.3m
behind Toyota Yaris
47

Simulation Example: p Collision
Car-to-Car Rear Case 3
� Ub Urban road denvironment: i t
– T-junction, daylight, fine
�� Honda Accord (striking):
– Following too close
behind BMW
– Travel speed 50km/h
– Impact speed 45km/h
� BMW 3 Series (struck):
– Travel speed 50km/h
– Impact speed 28km/h
� Ford Focus & Audi A4
(secondary):
– Travel speed 45km/h
– Impact speed 0km/h
48

Simulation Example: p AEB
Car-to-Car Rear Case 3
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds at 0.4g
– Activation at time to
collision lli i (TTC) 0.6 0 6
seconds at 0.8g
� Honda Accord
(striking):
– Travel speed 50km/h
� BMW 3 Series (struck):
– Travel speed 50km/h
– BBraking ki 00.35g 35
� Honda stops 0.3m
behind BMW
49

AEB Test Concepts p
Car to Car Rear (CCR)
� Car drives longitudinally
toward rear of proceeding
car which is stationary
� Car drives longitudinally
toward rear of proceeding
car which is travelling at a
slower speed
Provisional/DRAFT only
50

AEB Test Concepts p
Car-to-Car Rear (CCR)
� Car drives longitudinally
toward rear of proceeding
car which brakes
� Car drives toward rear of
proceeding car which is
stationary and obscured
around a bend
Provisional/DRAFT only
51

AEB Test Example p
Car-to-Car Rear (CCR)
� RRange of f ttest t speeds d from f 10 to t 60km/h 60k /h
� Speed increased in 10km/h increments if
system avoids collision with car target
� Speed increased in 5km/h increments if
collision occurs
� 100% overlap, possibly repeat test with smaller
overlap
� Goal – obtain AEB system performance curve
at multiple impact positions over a range of
speeds to quantify system performance
� Then perform additional scenarios at a reduced
number of speeds to investigate system
robustness
Provisional/DRAFT only
52

Test Target g and Propulsion p
Car-to-Car Rear (CCR) initial development
53

Test target g and propulsion
p p
Car-to-Car Rear (CCR)
� Representative of typical passenger car
� 3 dimensional
� Appropriate whole body radar cross section
� Appropriate pp p pproperties p for vision systems y
– Body structure, windows, lights, license plate etc.
– Under body shadow
� TTravel l speed d up tto 80k 80km/h /h
� Control to follow defined path and modulate speed appropriately
� Instrumentation communicating with test vehicle and data acquisition
� Repeatedly impactable at speeds up to 50km/h
� Operational p for a day y of testing g or capable p of fast charge g to top-up p p
54

Test Target g and Propulsion p
Car Crash Target
55

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
56

Car-to-Car Head-On
� Accidentology survey of UK data complete for head-on
collisions
� Simulation examples provided
� Current vehicle technology gy does not address this type yp of
collision, but is expected in near future
� Car-to-Car Head-On test scenarios and procedure p will be
developed after completion of work on CP and CCR
– CHO planned p for 2012-2014
57

Accidentology gy
STATS 19 — Head-on clusters
Aim for 4-6
clusters
≥75% of cases
Green cells:
over-represented
99 5% in χ2 99.5% in χ test
Example
Cluster 1
• 30% of cases
• Lower speeds
• At junction
•Daylight yg
• Fine weather
• Veh A going ahead
• Veh B turning
58

Accidentology gy
OTS — Accident Configuration and type
59

Accidentology gy
OTS — Head-on clusters
60

Simulation Example: p Collision
Car-to-Car Head-On Case 1
� RRural l road d
environment:
– Left hand bend bend,
daylight, fine
� Renault Clio (striking):
– Speeding on bend,
drifts out of lane
– Travel speed 65km/h
– Impact speed 40km/h
� Vauxhall Astra (struck):
– Travel speed 30km/h
– Impact speed 18km/h
61

Simulation Example: p AEB
Car-to-Car Head-On Case 1
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.3
seconds (this is the
time before the actual
impact)
� Renault Clio (striking):
– Travel speed 65km/h
– Braking 0.8g
– Impact speed 28km/h
� Vauxhall Astra (struck):
– Travel speed 30km/h
– Braking 0.3g
– Impact speed 18km/h
�� Mitigation of impact
severity
62

Simulation Example: p Collision
Car-to-Car Head-On Case 2
� Ub Urban road d
environment:
– Daylight Daylight, fine
� VW Golf (striking):
– Overtaking g Mazda
– Travel speed 75km/h
– Impact speed 54km/h
� Nissan X-Trail (struck):
– Travel speed 50km/h
– IImpact t speed d 31km/h 31k /h
� Mazda 6 (overtaken):
– Travel speed 40km/h
– Impact speed 34km/h
63

Simulation Example: p AEB
Car-to-Car Head-On Case 2
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.0
seconds (this is the
time before the actual
impact)
� VW Golf (striking):
– Travel speed 65km/h
– Braking 0.8g
– Impact speed 34km/h
� Nissan X-Trail (struck):
– Travel speed 50km/h
– Braking 0.35g
– Impact speed 29km/h
�� Mitigation of impact
severity
64

Simulation Example: p Collision
Car-to-Car Head-On Case 3
� RRural l road d
environment:
– Dark Dark, raining
� Ford Ka (striking):
– Lost control on bend
at speed and crossed
carriageway
– Travel speed 98km/h
– Impact speed 48km/h
� Audi A3 (struck):
– Travel speed 50km/h
– Impact speed 30km/h
65

Simulation Example: p AEB
Car-to-Car Head-On Case 3
� AAssume AEB AEB fitt fitted: d
– Activation at time to
collision (TTC) 1.6
seconds at 0.4g
– Activation at time to
collision lli i (TTC) 0.6 0 6
seconds at 0.8g
� Ford Ka (striking):
– Travel speed 98km/h
– Impact speed 33km/h
� Audi A3 (struck):
– Travel speed 50km/h
– Braking 0.3g
– Impact speed 30km/h
�� Mitigation of impact
severity
66

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
67

Metrics & Ratings g
� Wh Whether th or not t the th collision lli i was avoided id d
� The speed reduction prior to collision
� When (if present) a warning is given
� How an appropriate driver braking response at the time
of f warning i (if present) t) could ld have h affected ff t d the th outcome t
� Identify system performance curve for avoidance and
mitigation
� Correlate with injury risk curves to estimate potential
bbenefit fit
68

Agenda g
� AEB Test T t Group G
� CCar-to-Pedestrian t P d t i (CP) (CP):
– Accidentology
– Test Scenarios
– Test Target & Propulsion
� Car-to-Car Rear (CCR):
– Accidentology
– Test Scenarios
– Test Target g & Propulsion p
� Car-to-Car Head-On (CHO):
– Accidentology
� Metrics & Ratings
� Conclusions
69

Conclusions
� AEB Test Group formed to address the need for AEB
test procedures
� Small dynamic y ggroup p initially, y with consultation and
feedback from industry
� Representing p g consumer and insurer needs
� Collaborating with other projects (vFSS, ASSESS etc.)
– Development of pedestrian and car test targets
– Soft Crash Target Vehicle?
– Robust pedestrian?
� Euro NCAP road map for inclusion of AEB test within
vehicle rating system in 2013
� AEB Test Group output defining reference
70

Conclusions
� Test procedures developed from real world data
� Initial development tests undertaken proving concept and
assessment
� Test concepts published early 2011
� Finalise testing and development throughout 2011
� Undertake full tests in 2012 and publish p results
71

AEB
TTest t Development
D l t
www.thatcham.org/AEB