sip-adus workshop 2018...#17820 · 17zl00xx.pptx slide no. 3 2017/11/16 © fka 2018 · all rights...
TRANSCRIPT
-
© fka 2018 · All rights reserved2017/11/16Slide No. 1#17820 · 17zl00xx.pptx
Tokyo, 14th November 2018
Dr.-Ing. Adrian Zlocki, Christian Rösener, M.Sc., Univ.-Prof. Dr.-Ing. Lutz Eckstein
A Traffic-based Method for Safety Impact Assessment of Road Vehicle Automation
SIP-adus Workshop 2018
Forschungsgesellschaft Kraftfahrwesen mbH Aachen
-
© fka 2018 · All rights reserved2017/11/16Slide No. 2#17820 · 17zl00xx.pptx
MotivationPublic View on Automated Driving
► Research QuestionWhat is the safety level of automated driving?
► MethodologyA Traffic-based Method for Safety Impact Assessment of Road Vehicle Automation
How SAFE is Automated Driving?
-
© fka 2018 · All rights reserved2017/11/16Slide No. 3#17820 · 17zl00xx.pptx
I M P A C T A S S E S S M E N T
S A F E T Y A S S U R A N C E
Effectiveness
Controllability &Functional Safety
„Usability“
Pote
ntia
l con
flict
Acceptance Safety ofInteraction
Efficiency
Evaluation MethodologyImpact Assessment vs. Safety Assurance
SOTIFOperational Safety
Behavioral Safety
-
© fka 2018 · All rights reserved2017/11/16Slide No. 4#17820 · 17zl00xx.pptx
Analysis of Automated Driving Field Test DataScenario Classification of Real-World Data
FOTs
Par
amet
er x
Parameter z
Par
amet
er x
Parameter y
New generated situationsVariation of measured situations
Accident dataGeneration of new situations+
Variation of measured situations
Source: Eckstein, L., Zlocki, A.: Safety Potential of ADAS - Combined Methods for an Effective Evaluation, 23rd ESV 2013, Seoul, 2013
-
© fka 2018 · All rights reserved2017/11/16Slide No. 5#17820 · 17zl00xx.pptx
Impact Assessment of Automated DrivingDriving Scenarios from Accident Type
Overtaking oncarriageway
646 - Overtaking
631 – DeceleratingCut-In (right)
643 – Decelerating Cut-In (left)
Example: Passive Cut-In
-
© fka 2018 · All rights reserved2017/11/16Slide No. 6#17820 · 17zl00xx.pptx
Impact Assessment of Automated DrivingDriving Scenarios from Accident Type
Approach: The types of driving scenarios, respectively physicalaccident constellations, do not change with automated driving.
The frequency of occurrence and the severity of these drivingscenarios may change with automated driving.
-
© fka 2018 · All rights reserved2017/11/16Slide No. 7#17820 · 17zl00xx.pptx
Impact Assessment of Automated DrivingDefinition of Methodology for Impact Assessment
∆f(Sn)
AD
Ref
para
met
er i
parameter n
para
met
er i
parameter n
S1
S2
Sn
∆ Frequency of Occurrence
Scenarios
Derivation of Effectiveness Field
Automated driving function
∆ Severityin
Scenarios
Impact of automated driving function
Definition of Scenarios
∆ Frequency of occurence ofdriving scenarios
∆ Severity indriving scenario
Driving scenario-basedestimation of effectiveness field
1
2
3
4 5
6
3
45
-
© fka 2018 · All rights reserved2017/11/16Slide No. 8#17820 · 17zl00xx.pptx
Impact Assessment of Automated DrivingDefinition of automated driving functionand scenarios
200
m80
m 20°
60°
10 m
10 m
20°
20°
200
m
Fahr
trich
tung
Motorway-Chauffeur
Operational design domain (ODD):
Relevant driving scenarios:Driving
accidentApproachingstatic Object
Approachingvehicle
Approachinglateral Object
Automation level (SAE): 3
Traffic Jam
Cut-inLane change Take over
Turn at intersection
Crossingintersection Turn around
Operation domain:
12
21
-
© fka 2018 · All rights reserved2017/11/16Slide No. 9#17820 · 17zl00xx.pptx
v
Impact Assessment of Automated DrivingEffectiveness Field and Scenario Classification
FOTs
Par
amet
er x
Parameter z
Accident data
Scenario „Approaching vehicle“
Scenario „Lane change“Scenario
„Cut-in of other vehicle“
3
-
© fka 2018 · All rights reserved2017/11/16Slide No. 10#17820 · 17zl00xx.pptx
Driving scenario-based estimation of effective. fieldAccidents in Germany according to ODD
Federal motorway
6%
Federal highway
17%County road
14%
Rural road25%
Municipal road38%
Accidents in Germany in 2016308.145 A(P)
Accidents in domain „Motorway“19.010 A(P)
no car participation
11%
functional limits 17%
driver and vehicle related limits
11%
not adressable driving scenarios
14%
Effectiveness field 47%
3
-
© fka 2018 · All rights reserved2017/11/16Slide No. 11#17820 · 17zl00xx.pptx
Driving scenario-based estimation of effective. fieldInput data for scaling-up and simulation
Approaching traffic jam
18 %
Driving accident
17 %
Approaching leading vehicle
43 %
Passive cut-in16 %
Driving scenarios in effectiveness field
9.395 A(P)
Approaching static object2 %
Lane change4 %
Number of accidents forscaling-up of effectiveness
Situational variables foreffect of function by
simulation
Accidentstatistics
GIDAS
FOT
Driving scenario Accidents
Passive cut-in 1.219
v1=67 km/h
v2= 93 km/h
3
-
© fka 2018 · All rights reserved2017/11/16Slide No. 12#17820 · 17zl00xx.pptx
Safety Impact Assessment of Automated DrivingIdentification of ∆ frequencies of driving scenarios
∆f(Sn)
AD
Ref
∆ Frequencies of driving scenariosDriving scenario
based estimation ofeffectiveness field
Automated drivingfunction
∆Severityin drivingscenario
Impact of automateddriving function
Definition of drivingscenarios
∆Frequencyof drivingscenario
4
-
© fka 2018 · All rights reserved2017/11/16Slide No. 13#17820 · 17zl00xx.pptx
Identification of ∆ frequencies of driving scenariosFOT-data
Ref AD
Driving scenarioclassification
∆frequencies ofscenarios
© ika
AD
4
-
© fka 2018 · All rights reserved2017/11/16Slide No. 14#17820 · 17zl00xx.pptx
Impact Assessment of Automated Driving Identification of ∆ Frequency from FOT Data
AD
4
-
© fka 2018 · All rights reserved2017/11/16Slide No. 15#17820 · 17zl00xx.pptx
Identification of ∆ frequencies of driving scenariosTraffic simulation4
-
© fka 2018 · All rights reserved2017/11/16Slide No. 16#17820 · 17zl00xx.pptx
∆ Severity in driving scenarios by re-simulationSimulation framework
Simulation of reference
∆ Severity
Driving scenario„passive cut-in“
Ref
Simulationwith ADF
AD
► Human driverperformance modelsfrom driving simulator
study/FOT for reference
5
-
© fka 2018 · All rights reserved2017/11/16Slide No. 17#17820 · 17zl00xx.pptx
1% 3%5% 8% 4%
17%
30%
53%
4%
13%
24%
46%
4%
14%
26%
50%
3%
14%
26%
54%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
5% 25%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100%
Safety Impact Assessment of Automated DrivingImpact Assessment Results 6
Traffic Jam-Chauffeur
Motorway-Chauffeur
CommuterChauffeur
Universal-Chauffeur
Urban Robot-Taxi
Effe
ctiv
enes
s in
dom
ain
-
© fka 2018 · All rights reserved2017/11/16Slide No. 18#17820 · 17zl00xx.pptx
Safety Impact Assessment of Automated DrivingImpact Assessment Results 6
0
50,000
100,000
150,000
200,000
250,000
300,000
5% 25%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100% 5% 25
%
50%
100%
Acc
iden
tsw
ithpe
rson
al in
jurie
spe
r yea
r
accidents in domain, of themwithout involvement of passenger caroutside the functional limitsnone or ambiguous effectavoided
Traffic Jam-Chauffeur
Motorway-Chauffeur
CommuterChauffeur
Universal-Chauffeur
Urban Robot-Taxi
-
© fka 2018 · All rights reserved2017/11/16Slide No. 19#17820 · 17zl00xx.pptx
Safety Impact Assessment of Automated DrivingImpact Assessment Results
1% 3% 5% 8% 4% 17% 30% 53% 4% 13%24%
46%
4%
14%
26%
50%
3%
14%
26%
54%
0
50.000
100.000
150.000
200.000
250.000
300.000
5% 25%
50%
100% 5% 25
%50
%
100% 5% 25
%
50%
100% 5% 25
%
50%
100% 5% 25
%50
%10
0%
Acc
iden
ts w
ith p
erso
nal
inju
ries
per
yea
r
accidents in domain, of themwithout involvement of passenger caroutside the functional limitsnone or ambiguous effectavoided
effectiveness in domainX %
Motorway-Chauffeur
Traffic Jam-Chauffeur
Commuter-Chauffeur
Universal-Chauffeur
Urban Robot-Taxi
6
-
© fka 2018 · All rights reserved2017/11/16Slide No. 20#17820 · 17zl00xx.pptx
Safety Impact Assessment of Automated DrivingKey results
Motorway-Chauffeur can reduce 30 % of all accidents on German motorways at a market penetration of 50 %. This equals 2 % of all accidents on German roads.
The Urban Robot-Taxi can avoid 26 % of all accidents with personal injury within city-limits at a market penetration of 50 %. This equals 17 % of all accidents on German roads.
However, there will be accidents remaining that automated vehicles cannot avoid (due to weather conditions or physics). But we can show that a human cannot avoid these accidents either.
AD Ref
6
-
© fka 2018 · All rights reserved2017/11/16Slide No. 21#17820 · 17zl00xx.pptx
Piloting Automated Driving on European RoadsL3Pilot – Real World Data for Impact Assessment
Large-scale Level 3 piloting 1,000 test drivers,100 vehicles
in 11 European countries EC funded in Horizon 2020 34 partner Budget: 68 € Mio., Funding: 36 € Mio.
Website: http://www.l3pilot.eu
-
© fka 2018 · All rights reserved2017/11/16Slide No. 22#17820 · 17zl00xx.pptx
L3Pilot Evaluation Levels
Data Management
Technical & Traffic Evaluation
Socio-economic impact evaluation
User evaluation
Impact EvaluationSafety impact (e.g. avoided accidents)Environmental impact (e.g. fuel consumption)on European target level
Objective driving data
Europe
Subjective data
∆ Frequency of driving scenariosDMs, PIs per driving scenarios
e.g. Effect in Transition of controlLong-term effects, usage
-
© fka 2018 · All rights reserved2017/11/16Slide No. 23#17820 · 17zl00xx.pptx
L3PilotEvaluation Workflow
23
Technical & Traffic Evaluation
Vehicles – Obj. Data Subj. Data
Digital versionCDF Time series data
User & AcceptanceEvaluation
Impact Assessment
Estimated Impacts
Socio-Economic Impact Assessment
Impact Assessment
Socio-Economic Impact
Technical Evaluation
User Evaluation
Other inputs(e.g. accident & baseline data)
Performance Indicators per driving scenario
-
© fka 2018 · All rights reserved2017/11/16Slide No. 24#17820 · 17zl00xx.pptx
Summary
Prospective safety impact assessment for automated driving requires new methodologies
Automated driving provides many challenges with regards to impact assessment since limited real world data is available yet and many new aspects (e.g. user-interaction) needs to be taken into account
Safety impact assessment shows positive results with different automation function
Current research in L3Pilot start data collection for safety impact assessment
Safety Impact Assessment in L3Pilot will provide results based on data from vehicles combined with simulation for the first time
-
© fka 2018 · All rights reserved2017/11/16Slide No. 25#17820 · 17zl00xx.pptx
PhoneFax
EmailInternet www.fka.de
fka Forschungsgesellschaft Kraftfahrwesen mbH AachenSteinbachstr. 752074 AachenGermany
Contact
Dr.-Ing. Adrian Zlocki
+49 241 80 25616+49 241 8861 110
THANK YOU FOR YOUR ATTENTION!
QUESTIONS?