Pedestrian acoustic warning system for electric vehicle
PBNv2 – H20202 Marie Curie ETNFirst Public Technical Course; 28th – 29th November 2018
INSA - Lyon
PUBLIC INTERNAL CONFIDENTIAL SECRET
Quiet vehicle audible alert systems: current situat ion
The recent rapid growth of very quiet Electric Vehicles and Hybrids has prompted new concerns about pedestrian safety.
The solution proposed by vehicle manufacturers and legislators is to add synthetic vehicle sounds to EV’s (EC/EV 540/2014)
The objective is simple to make quiet EV’s detectable by pedestrians through the use audible cues.
But is this solution really that simple to implemen t, to everyone’s satisfaction?
• Some fundamentals generally accepted:
• Pedestrians and Vulnerable Road Users need to be aware of the presence of any vehicle which could pose a collision risk, in time to take avoidance action.
• The impact on environmental noise pollution should be minimised
• The synthetic alert sounds added to cars should not result in customer rejection of EV’s and HEV’s
Quiet vehicle audible alert systems: restrictions
Country
Spain
CategoryLeaders of Work PackagesConsortium Member Organisations
-
* WP4
-
* WP1
* WP6
* WP5
* WP2
WP3
-
* WP11
* WP7, * WP8, WP9 & WP10IDIADA Automotive Technology IDIADA
R&DCentre
LMS International NV LMS Belgium
ÖSTERREICHISCHES FORSCHUNGS- UND PRÜFZENTRUM ARSENAL
AIT Austria
TOEGEPAST NATUURWETENSCHAPPELIJK ONDERZOEK TNO Netherlands
Institut National des Sciences Appliquées INSA-Lyon FranceUniversities
Technische Universitaet Darmstadt TUD Germany
NISSAN Motor Manufacturing (UK) Nissan United Kingdom
OEM’sRenault Renault France
PSA PSA France
Continental Continental France Tier-1’s
European Blind Union EBU France End Users
Country
Spain
CategoryLeaders of Work PackagesConsortium Member Organisations
-
* WP4
-
* WP1
* WP6
* WP5
* WP2
WP3
-
* WP11
* WP7, * WP8, WP9 & WP10IDIADA Automotive Technology IDIADA
R&DCentre
LMS International NV LMS Belgium
ÖSTERREICHISCHES FORSCHUNGS- UND PRÜFZENTRUM ARSENAL
AIT Austria
TOEGEPAST NATUURWETENSCHAPPELIJK ONDERZOEK TNO Netherlands
Institut National des Sciences Appliquées INSA-Lyon FranceUniversities
Technische Universitaet Darmstadt TUD Germany
NISSAN Motor Manufacturing (UK) Nissan United Kingdom
OEM’sRenault Renault France
PSA PSA France
Continental Continental France Tier-1’s
European Blind Union EBU France End Users
eVADER was a 3 year project, which kicked-off in October 2011 and was finished in
December 2014 (Electric Vehicle Alert for Detection and Emergency Response)
The consortium members were:
eVADER - Consortium Members
eVADER proposed an ‘smart’ audible alert system, to find an optimum balance between pedestrian safety in the vicinity of EV’s and minimising environmental noise pollution, but also securing customer acceptance.
eVADER technical direction - Concept
Default condition broad beam at lower sound level
Alert sound directed toward pedestrian at increased sound level, dependant on risk estimation
The pedestrian alert sound was emitted by a Beamforming Loudspeaker Array (directed towards the detected VRU – Vulnerable Road User) The sound beam could optionally be rotated from one side of the vehicle to the other side within ~0.5s if required.
eVADER technical direction - Concept
Pedestrian detection by EPS camera system
Tracing pedestrians paths for better risk estimations
eVADER incorporates an Environmental Perception System (EPS) based on existing ADAS [Advanced Driver Assistance Systems] able to detect any Vunerable Road User (VRU)The EPS will be coupled with a Location Based System (LBS ) which will give GPSposition, time of day and a database of hotspots, critical areas and speed limits, to support risk estimations when no VRU is detected.
eVADER technical direction - Pedestrian
eVADER pedestrian risk estimation concept is divided into two main topics. • Risk based on location / time (LBS)• Risk based on pedestrian detection (EPS), risk estimator output = collision
probability and time to collision.
eVADER – System Overview
Stereo CameraMFC310
FrontRadar
ARS351
Vehicle to CONTI
CANGateway
EberspacherFlexconMidget
CONTI CAN
CONTINENTAL
LBS (Squarell GPS)
Pedestrian detection & LBS system hardware
Can Bus Router & Vsound DSP Unit
Risk Estimator & Interaction Manager
F2016-ETNVH-1
Stereo camera - Installation
Special windscreen used with masking profile to match Pedestrian Detection Camera
T1018_full.avi T1032_full.aviT1033_full.avi T1034_full.avi
Windscreen maskingprofile
Mic 1 Mic 2
Mic 3 Mic 4
Objective
eVADER system monitors streetambient sound levels to allow forcompensation in the alert sound levelto help audibility.
Selected microphone
Knowles Sisonic microphone (MEMS):
Low cost
Robust, tolerant of environmentalconditions and vibration.
Microphone array design
Three main aspects of vehicle sound were studied by eVADER
1: Detectability and the influence of three timbre parameters: - Frequency content - Frequency modulation- Temporal modulation
2: Ability to judge vehicle speed / distance from a vehicle sounds only3: Annoyance of warning sounds
eVADER explored the balance between good detectability of EV sounds, minimum environmental noise annoyance and customer acceptance.
eVADER psychoacoustics
Experiment 111 stimuli (Diesel car, EV, EV+Alert sound).8 repetitions for each stimuli (left-right or vice-versa).Background noise (69 dBA).Headphone presentation (Stax Lambda Pro).
Task :Detect the car as soon as possible, identify the di rection of the car (left/right).112 participants (36 visually impaired, 86 sighted).
eVADER psychoacoustics – 1. Detectability
"Waiting-to-cross" scenario passing car (20 km/h)
- 30 m 30 m
Recording of a diesel and an electric vehicle Synthesis of the 9 alert sounds Binaural simulation of a moving sound sourceMixing with EV recording
eVADER psychoacoustics – 1. Detectability
88 trials for each participant : averaged response time was computed
0
2
4
6
8
10
12
14
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Distance at detection
Num
ber
of p
artic
ipan
ts
Key Finding:No difference between Visually-Impaired and Sighted subjects.
eVADER psychoacoustics – 1. Detectability
Distance for detection
Key Finding:Some alert sounds were as easy to detect as the diesel car
eVADER psychoacoustics – 1. Detectability
F2016-ETNVH-1
30
25
20
15
10
5
0
111 122 133 212 223 231 313 321 332 EV Dieselstimulus
dist
ance
to th
e pe
dest
rian
(m)
Key finding: No relationship to sound level
64
66
68
70
72
74
76
78
111 122 133 212 223 231 313 321 332 Electric Diesel
stimulus
dB(A
)Maximum ofA-weighted level
Detectability
eVADER psychoacoustics – 1. Detectability
Low detectability
-6
-4
-2
0
2
4
6
0 1 2 3 4
level
effe
ct (m
) F1F2F3
Factor DescriptionLevel
L1
Level
L2
Level
L3
F1No. of
frequencycomponents
3 6 9
F2Frequency
mod.None Sine
Saw
tooth
F3Temporal
mod.None Sine Chaotic
Key Finding: Timbre parameters which aid detection• small frequency range• Temporal modulation
Influence of timbre parameters:
eVADER psychoacoustics – 1. Detectability
F2016-ETNVH-1
Procedure for experiment 2
Same "waiting-to-cross" scenario
Task : Indicate when you think it is no longer safe to cross the road while this car isapproaching.
20 Stimuli :9 alert sounds2 speeds (20 km/h, 30 km/h).Diesel car included.Background noise (64 dBA).4 repetitions.
116 participants (39 Visually Impaired, 77 Sighted).
- 30 m 30 m
Key Finding :- Visually Impaired participants were on average 0.5 sec quicker to respond to approaching car sounds
eVADER psychoacoustics – 2. Speed / distance
-400
-300
-200
-100
0
100
200
300
400
low mid high
effe
ct (m
s)
pitch
mod. freq.
speed
Key Findings:- Speed effect as expected: People feel unsafe sooner for cars at
30 km/h compared to 20 km/h.- No effect of modulation frequency.- Surprising effect of pitch: higher pitch tended towards later
judgement of unsafe to cross.
Factors effect : Pitch, Modulation freq. & Vehicle speed
eVADER psychoacoustics – 2. Speed / distance
All sounds were as used in previous experiments
Procedure :Sounds corresponded to a car passing in front of the listener at 20 km/h – with nobackground noise.Each sound presented twice, participants could listen to each sound as often asthey wanted before moving on.
Task : Listen to the sound then move the slider to indicate how unpleasant thesound was(scale: not at all unpleasant - to - extremely unpleasant)
Key Findings : - The EV was judged the least unpleasant sound. - Four of the synthetic sounds judged as equally unpleasant as the diesel – all
others were judged more unpleasant
eVADER psychoacoustics – 3. Annoyance
Factors effect : Frequency modulation, Number of harmonics & Temporal modulation
Key Findings :- Temporal fluctuation increases unpleasantness- Increasing number of harmonics tends to lower unpleasantness
eVADER psychoacoustics – 3. Annoyance
MoreAnnoyance
LessAnnoyance
Distance-to-detection chart, highlighting the synthetic sounds with lowest annoyance.
Sound 313 – a good compromise between detectability and annoyance which is also 6dBA quieter than the diesel.
eVADER psychoacoustics – 3. Annoyance
System requirements:- Minimise number of acoustic sources (6)- Frequency range 300Hz to 1.2kHz- SPL up to 90dBA at 1 metre- Directivity only required in horizontal plane- Beam steering nominally ±60 degrees- Angular tracking speed ~300 deg/sec- Non uniform loudspeaker array
A range of potential beamforming algorithms have been evaluated. The ‘sound power minimization’ algorithm was chosen for eVADER as it provides narrow beams and allows flexible real-time implementation.
eVADER Beamforming loudspeaker array
Simulation result, 30 degree orientated alert signal spatial distribution at 600Hz
Simulation model forLoudspeaker array
Example: Beamforming Directivity performanceas a function of frequency for a 30 degreeorientated alert signal at distance of 20m fromthe bumper loudspeaker array.
50mm Speakers
Evaluation of beamforming system performance using FEM & BEM modelling techniques. Influence of ambient temperature, humidity, ground reflections and beam scattering from parked vehicles have been considered.
eVADER Beamforming loudspeaker array
Bumper & loudspeaker test system prepared by Technical University of Darmstadt –using CAD data supplied by Nissan
eVADER Beamforming loudspeaker array
The eVADER pedestrian alert system was implemented into a Nissan LEAF vehicle for evaluation and testing.
Systems for integration:• External microphones• EPS system, camera & radar• LBS system, GPS• Risk Estimator computer• Beamforming loudspeaker array• Beamforming controller and internal
alert sound generator• Communications on dedicated
eVADER CAN bus
eVADER System integration into vehicle
LEAF
Mic 1 Mic 2
Mic 3 Mic 4
AUX in
Speaker Array
Speaker Amps
EVADER CAN BUS HS 500KPCAN ProRouter
MicPre-Amps
V-CAN1
Riskestimation
Ped.filtering
Driverbehavior
Interactionmanager
LBS algorithms
Environmentalnoise level
Environmentalnoise level
CAN bus data
Processing
LMS Stereospeaker TFs
Exterior sounddatabase
Interior sounddatabase
BKSVSound algorithm
Exterior
Interior
TNOspeaker
beam-forminglibrary
1-6
LineOut
7-8
NissanDSP V Sound
TNO Craneboard
V SoundRP feedback
Stereo CameraMFC310
FrontRadar
ARS351
Vehicle to CONTI
CANGateway
EberspacherFlexconMidget
CONTI CAN
CONTINENTAL
LBS (Squarell GPS)
eVADER System integration into vehicle
Assessment : Warning sound detectability
Movement of jury member
Pointof pedestrian
detection
Measurement point
SPL (dB(A))
Loudness(Phone)
Loudness(Sone)
1 1 62.0 70.1 8.12 2 48.5 59,3 3.83 3 58.5 67,0 6.54 4 46.9 57,6 3.45 5 57.2 65,9 6.06 6 45.3 57,1 3.36 4 33.2 44,4 1.4
Assessment : Detectability
Assessment : Beam forming & detectability
Assessment: Polar plots of beam former
Vehicle turning (Near – Pos 1)Vehicle turning (Away) – Pos. 1)
Assessment : In a real urban environment
Vehicle turning (Near) – Pos 2
Vehicle turning (Near) – Pos 2
Assessment : In a real urban environment
Walking to backing vehicle (Near) Walking to backing vehicle (Away)
Waiting to cross (Near) Waiting to cross (Away)
Assessment : Pass by test configuration
10 dB(A) difference between risk
and non-risk area
Questions about eVADER’s performance Global % VIP % Non -VIP %Yes No Yes No Yes No
1)Do you feel that eVADER sound exceeds that of a typ ical conventional internal combustion vehicle in similar operating conditions?
0 100 0 100 0 100
2)Do you associate the eVADER sound with a vehicle? 54 46 60 40 50 503) Does eVADER inform you of the vehicle location in the street?
86 14 73 27 95 5
4) Does eVADER’s sound inform pedestrians of the veh icle motion (approaching, stationary, etc.)
75 25 67 33 82 18
5) What do you think about eVADER’s sound characteristics? ( 5 = Very good / 1 = Poor ) 3.7 3.3 4.0
6) Do you find eVADER’s sound is acceptable for its purpose?
84 16 67 33 95 5
7) Do you find eVADER’s sound detectable in ‘at risk ’ traffic situations with pedestrians?
70 30 40 60 91 9
Assessment : In a real urban environment
The main objective of the eVADER project was to develop a next generation alert system solution which will resolve as much as possible the conflicting requirements of minimizing the impact of EV sounds on traffic noise pollution levels, whilst providing effective audible cues to alert pedestrians and other vulnerable road users of the presence of a nominally silent EV /HEV.
The contribution of eVADER technology to the potential reduction of accidents caused by HEV /EV involving pedestrians has been assessed using basic considerations regarding the relative percentage of EV and the level of implementation of eVADER technology to these EV.
The results show that the contribution of eVADER technology to the reduction of accidents involving HEV/EV and vulnerable road users is potentially high, with negligible impact on urban noise pollution.
Conclusions
This work has been part funded by the European Commission under the 7th Framework Programme as a collaborative project FP/-SST-2011-RTD-1 with Grant Agreement Number 285095.
Thanks are due to all the consortium members for their contributions in order to materialisethe ideas presented in this work on a real vehicle demonstrator: INSA-Lyon, TNO, LMS, AIT, Continental, Nissan NTCE, Renault, PSA, TUD, The European Blind Union, plus the support of B&K.
eVADER - Acknowledgements
For further information:
Email: [email protected] IDIADAHeadquarters & Main Technical CentreT +34 977 166 000 (Barcelona)e-mail: [email protected]
Applus IDIADA
Headquarters & Main Technical CentreT +34 977 166 000 (Barcelona)e-mail: [email protected]
Applus IDIADA BelgiumT +32 2 719 02 45 (Brussels)e-mail: [email protected]
Applus IDIADA Brazil T +55 41 3373 0411 (Curitiba) T +55 11 4330 9880 (São Paulo) T +55 15 3205 2952 (Tatuí) e-mail: [email protected]
Applus IDIADA ChinaT +86 10 8446 3317 (Beijing) T +86 431 8190 9680 (Changchun) T+ 86 23 6756 8060 (Chongqing) T +86 20 2282 9202 (Guangzhou) T +86 (21) 6210 0894 (Shanghai) T +86 (772) 3166 619 (Liuzhou)T +86 (755) 29184532 (Shenzhen)e-mail: [email protected]
Applus IDIADA Czech RepublicT +420 493 654 811 (Hradec Králové) T +420 482 424 243 (Liberec) T +420 326 736 860 (Mladá Boleslav)e-mail: [email protected]
Applus IDIADA MalaysiaT +603 9207 7018 (Kuala Lumpur)e-mail: [email protected]
Applus IDIADA Mexico T +52 (1) 222 170 6722 (Puebla)e-mail: [email protected]
Applus IDIADA PolandT +48 61 6226 905 (Poznan)e-mail: [email protected]
Applus IDIADA RussiaT +7 (831) 297 94 32 (Nizhny Novgorod)T +7 (831) 261 37 06 (Togliatti)e-mail: [email protected]
Applus IDIADA Saudi ArabiaT +966 53 4147 301 (Riyadh)e-mail: [email protected]
Applus IDIADA South AfricaT +27 83 450 8925 (Pretoria)e-mail: [email protected]
Applus IDIADA South KoreaT +82 31 478 1821 (Seoul)e-mail: [email protected]
Applus IDIADA FranceT +33 (0) 141 146 085 (Sèvres)e-mail: [email protected]
Applus IDIADA India T +91 986 0098 659 (Bangalore) T +91 994 0679 933 (Chennai) T +91 124 4028 888 (New Delhi) T +91 203 0556 900 (Pune)e-mail: [email protected]
Applus IDIADA Indonesia T +6221 2939 1143 (Jakarta)e-mail: [email protected]
Applus IDIADA ItalyT +39 011 016 0205 (Turin / Maranello)e-mail: [email protected]
Applus IDIADA JapanT +81 (0) 42 512 8982 (Tokyo)e-mail: [email protected]
Applus IDIADA MadridT +34 915 095 795 (Madrid)e-mail: [email protected]
Applus IDIADA TaiwanT +886 47 810 702 (Lu-Kang)e-mail: [email protected]
Applus IDIADA ThailandT +66 86 7917 071 (Bangkok)e-mail: [email protected]
Applus IDIADA TurkeyT +90 216 250 6050 (Istanbul)e-mail: [email protected]
Applus IDIADA UKT +44 2476 328 083 (Nuneaton)e-mail: [email protected]
Applus IDIADA USAT +1 248 978 0111 (Detroit)e-mail: [email protected]
CTAG IDIADA T +34 986 900 300 (Pontevedra)e-mail: [email protected]
Thank you !Questions ?