Vous êtes sur la page 1sur 14

Frontal Collision Warning Sound

Assignment 3

0P900 Automotive Human Factors

K. Bouwman 0667412
A. Kaipencheri 0926274
K. Mundaragi 0926061
M. K. Soundarapandian 0925101
J. Wijkniet 0871034

Coordinator: Dr. D.J. Hermes

Date: 3 October 2014


Content

1. Introduction ................................................................................................................... 2
1.1 Assignment.................................................................................................................................... 2
1.2 Audio Alert Requirements: ........................................................................................................... 3

2. Studies of Automotive Cabin noise.................................................................................. 5


2.1 Study by ‘Autobild’ - 2011............................................................................................................. 5

3. Design Principles ............................................................................................................ 6


3.1 Detectability, Audibility................................................................................................................. 6
3.2 Identifiable .................................................................................................................................... 6
3.3 Discriminability ............................................................................................................................. 6
3.4 Localizability .................................................................................................................................. 6
3.5 Perceived Urgency ........................................................................................................................ 6
3.6 Annoyance .................................................................................................................................... 7
3.7 Timing............................................................................................................................................ 7
3.8 False alarm .................................................................................................................................... 7
3.9 Number ......................................................................................................................................... 7
3.10 Startle Responses .......................................................................................................................... 7

4. Design decisions ............................................................................................................. 8


4.1 Intensity and pitch ........................................................................................................................ 8
4.3 Warning sound ............................................................................................................................ 10

5. References ................................................................................................................... 11

6. Appendix A. Matlab Script ............................................................................................ 12

Page 1
1. Introduction
Auditory warning systems are being used in a car for multiple situations. Doors ajar, headlamps
on under key-out, indication of low fuel, crossing a set speed limit, seatbelt warnings, parking
brake not set free, park distance control, remote locking, theft alarms and even to notify faults
in an engine management system. The list of auditory alerts in a car is long. However it is
interesting to note that the type of auditory alerts is different for each of the above situations.

From experience any driver of a modern car will agree that he can easily distinguish between
audio alerts of low fuel warning, a park brake not set free to a park distance control. This is an
effective consideration of auditory human factors where the situational urgency is factored in
with varying envelope, intensity, rhythm and pitch of the audio alert.

1.1 Assignment
In conjunction with developing a system that warns the driver when safe distances are not
maintained with the vehicle moving ahead, an auditory alert is proposed when the distances
between the moving cars goes into the red level (as per the visual model submitted earlier)

This assignment deals with the design and synthesize of an auditory frontal-collision warning in
Matlab. In the design the following problems are considered:

 Detectability/Audibility;
 Identifiable;
 Discriminability;
 Localizability;
 Perceived urgency;
 Annoyance;
 Timing;
 False alarms;
 Number;
 Startle responses;

Page 2
1.2 Audio Alert Requirements:

The audio alert system will work as a sub-function of the safe distance monitory system and the
visual interface modeled for it. Hence the trigger points for the audio alert are based on the
same inputs used for the visual system earlier (in this case the red car of the visual interface).

The following additional considerations are made for the audio alert:

The audio alert should be heard above the ambient noise level which includes:
 Alarm should be heard above the background ambient noise and so it should be tailored
to be at least 15 dB above the threshold of hearing above the noise level. This typically
requires about 30-dB difference above the noise level in order to guarantee detection
 The engine noise of the vehicle as perceived from the cabin. Assumption is made that
vehicle is being driven with windows rolled up
 Road noise – due to the contact between road and tyres. Variations based on speed,
type of tyre, road surface material (asphalt, concrete etc.) exist.
 The stereo/music system that maybe operational in the car – A facility to automatically
minimize the stereo sound when the safe distance audio alert is being played
 Conversation of passengers inside the car (among each other , over a telephone, hands-
free calling using the car’s speakers)
 Other noises from the environment (under highway conditions) – noise from other
vehicles, weather deterioration (rain/thunderstorms), passing through tunnels

The safe distance audio alert should clearly distinguish itself from other audio alerts discussed
above
 The alert should be perceived by the driver as more critical than a low fuel / door ajar
alert
 The alert shouldn’t be as intense or startling as a Park distance control (when the
obstacle is very close by)
 Depending upon how long the driver remains in ‘red zone’, the alert will slowly vary its
intensity and urgency (by varying envelope, pulses and pause widths).
 The above can be function of whether the distance (though unsafe) is being maintained
constantly or if the gap is being closed quickly, impending a possible crash.
 The audio alert will follow an ‘initial detection’ – ‘identification’ – ‘driver alerted’ stages
by varying the intensity envelope, pause width and pulses.

Page 3
 The alert should not be overly abrupt or startling. A panic situation that will cloud the
driver’s judgment should be avoided in highway situation.
 The alert should not disrupt the understanding of other signals – In the current context,
this may be a GPS navigation instruction, a major fault in engine (overheating / loss of
oil pressure, etc).
 The alert should be informative – the reaction time of driver to the alert should be
minimal and it should not cloud his absolute judgment.

Page 4
2. Studies of Automotive Cabin noise
2.1 Study by ‘Autobild’ - 2011
Below is the tabular extract from a study of cabin noise of various passenger car conducted in
2011 by the German Automagazine AutoBild. The study acknowledges the fact that actual
perception of loudness will be factored more by the environmental conditions. Also only a
simple decibel meter held by hand at various locations was used for the measurements.

Name of car Noise level in dB

BMW 730d Blue Performance 58


Jaguar XJ 3.0 V6 Diesel S Luxury 58
Audi A8 4.2 TDI DPF quattrotiptronic 59
VW Touareg 4.2 V8 TDI DPF Automatik 59
Jaguar XF 3.0 V6 Diesel 60
Audi Q7 3.0 TDI DPF quattrotiptronic 60
BMW X6 ActiveHybrid 60
Mercedes R 350 CDI L DPF 4Matic 7G-Tronic 60
BMW 520d Touring 61
Mercedes S 400 Hybrid 7G-Tronic 61
VW Sharan 2.0 TDI BlueMotionComfortline 61
Audi A4 2.0 TDI DPF Ambition 61
Saab 9-5 2.0T Aero 61
Porsche Cayenne S Hybrid Tiptronic 62
Mercedes-Benz E 200 CGI BlueEfficiencyAvantgarde 62
Volvo S60 D3 Summum 62
BMW 135i CabrioAutomatik 62

Source: http://elevatingsound.com/the-top-30-quietest-cars-a-cabin-noise-test-by-auto-bild/

From the above study, we can assume a cabin noise value of 60dB for designing our interface.

Page 5
3. Design Principles
When designing an alarm sound, the following guidelines should be considered:

3.1 Detectability, Audibility


In a car there are many different sounds and noises, such as wind noise, tire noise and human
communication as well as other sounds from the music system, navigation system etc. The
warning sound should rise above all this background noise and be heard clearly by the user. To
prevent the sound from being masked by other noises, the alarm signal should consist out of
different frequencies which intensity should be at least 15dB above the intensity of the
background noises, although this value may be larger if the pitch of the sound to be heard is
unknown. Low-pitch sounds mask high-pitch sounds more than the converse. The alarm should
not be above the danger level for hearing, whenever this condition can be avoided.

3.2 Identifiable
The driver has to know which system alerts him with the warning sound. If it takes several
seconds before the driver identifies the sound and the corresponding danger, it might be too
late to respond to the oncoming danger. It is for this very reason the warning sound for this
system needs to be unique. The alarm sound should be synthesized in the way that it sounds
like the condition it represents, called auditory icons or earcons.

3.3 Discriminability
To guarantee informative and to minimize confusability, designers should try to stay within the
limits of absolute judgments. Discriminability is closely related with identifiable nature of the
sound. The driver should be able to distinguish between several sounds.

3.4 Localizability
The sound has to come from the direction where the danger exist. In case of a frontal collision
danger, the danger is in front of the car. The perceived location of the sound by the driver
should therefore be from the front and emanate closer to the dash board of the vehicle.

3.5 Perceived Urgency


The alarm should be informative and it helps in communicating the urgency even when the
driver is not visually alert. A frontal collision is a huge danger. The warning sound for this
system should therefore be received which immediately alerts the driver’s attention. The
timbre, rhythm and pitch of the warning sound determine the perceived urgency and the
driver’s response to the alarm.

Page 6
3.6 Annoyance
Annoyance is inverse related to perceived urgency. To make sure the signal is not unduly
annoying; it’s not advisable to choose the sound intensity that is too high. Also the starting
pulses can be configured with a rise envelope that is not too abrupt so that it will avoid the
"startle" created by more abrupt rises.

3.7 Timing
Timing of warning sounds is very important. The timing of alarm sound should be made perfect,
so that the intended user has enough time to react to the problem.

3.8 False alarm


False alarms can make the users to reduce the system usage over a period of time. False alarms
can be mitigated by providing additional warning signals through other sensory media. This can
be done for instance via a visual display or force feedback (vibration).

3.9 Number
Number and placement of alarm should be engineered so as to avoid any kind of confusion of
the source of sound.

3.10 Startle Responses


A person’s initial response to any alarm is startle followed by panic. If the driver starts to panic
and pushed the throttle, the situation can become even more dangerous. This can be done by
ensuring that the intensity isn’t too high and/or increasing the intensity and loudness of the
beeps in steps. This prepares the driver for the high intensity beep which is needed for
overcoming the masking noise problem.

Page 7
4. Design decisions

In this chapter the specific design choices that were made are explained, and related to the ten
principle criteria for a good warning sound.

4.1 Intensity and pitch


The average intensity of the alarm signal is chosen around 75dB, which is 15 dB higher than the
average background noise of 60dB of a modern car, see Paragraph 2.1. This ensures the driver
is alarmed by the sound, but not scared by the intensity.
In order to avoid startle response the warning sound consists of two tones after each other. The
first tone uses an amplitude 65 dB and a frequency range of 500 to 2500Hz. The second tone
uses an amplitude of 75 dB and a frequency range of 580 to 2900Hz. The chosen frequency
range is within the hearing range of the human 20 – 20.000 Hz and especially around the most
sensitive region of 2000 to 4000Hz. The small variation used in the warning sound amplitude
causes the driver to better discriminate it from the surrounding noise. The frequencies where a
signal has energy are related to the pitch and tone colour of the sounds. The amount of power
at each frequency (the power spectrum) can be calculated using the (fast) Fourier Transform
(FFT) [7]. The Fourier transformation of the sound is plotted in figure 4-1(right).

Figure 4-1 Audio amplitude as a function of time (left) and frequency spectrum of the signals.

The magnitude of the values in the audio signal is related to its perceived intensity or loudness
Loudness is measured as an average over time. In the script, the root mean square (RMS)
amplitude is calculated and converted to dB-scale. For higher frequencies, the amplitude is
reduced since humans are more sensitive for these frequencies.

Page 8
The average amplitudes of signal a and b correspond with 65 and 75 dB, respectively (see Figure
4-1 left). To prevent masking of the warning signal by vehicle frequencies, signal a as well as
signal b are composed of 5 harmonics. The fundamental frequency of signal a and b are 500 and
580 Hz, respectively. The harmonics contains the frequencies of the fundamental multiplied by
integer 2 to 5, sequentially. The period time (inverse frequency) can be seen in Figure 4-2

Figure 4-2 Signal shapes of fundamental frequencies and harmonics

The warning sound could easily irritate the driver, to prevent this warning signal sounds five
times and is then interrupted for three seconds and then repeats itself five times, etc. See
Figure 4.3. The signal of Figure 4-1 correspond with one warning signal as can be seen in Figure
4-3.
Amplitude

3sec

Figure 4-3: Schematic overview warning signal activation over time Time

Page 9
4.3 Warning sound
The warning sound is played when the red car appears on the interface screen. In that way
driver is besides visually as well audibly warned by his unsafe distance, see Figure 4.2.
The advantage of the warning sound compared with the visual warning is that it is
omnidirectional, this means the driver doesn’t have to take is attention of the road, but knows
by hearing the warning sound that he is to close. The Matlab script used to make the sound is
presented in Appendix A.

Figure 4.4: Interface display with warning LED’s active

Page 10
5. References
The following references have been used:

1. Introduction to Human Factors Engineering, Christopher D. Wickens (Author), John D. Lee


(Author), Yili Liu (Author), Sallie Gordon-Becker (Author)
2. Lecture Slides, Automotive Human Factors – TU/e
3. http://www.bksv.com/NewsEvents/Waves/Waves_2014_1/03-quietest-car -- An article about
measuring cabin noise of modern cars
4. http://elevatingsound.com/the-top-30-quietest-cars-a-cabin-noise-test-by-auto-bild/ - A 2011
cabin noise study – Not as accurate as the previous reference.
5. Hill,D.R., A Conceptionary for speech and hearing in the context of machines and
experimentation, The University of Calgary(2001).
6. 2David Bernstein., An Introduction to Auditory content, James Madison University (2004)
7. ELE595, Software Tools for Engineers, Lecture 4: Applications of MATLAB, Queen Mary
University of London

Page 11
6. Appendix A. Matlab Script
clear all; close all; clc

Fs=12000; % Sampling rate


T=1/Fs; % Period time
L=1500; % Length of signal
t= (0:L-1)*T;

fa_ground=500; % Fundamental frequency signal a


fb_ground=580; % Fundamental frequency signal b

x1 = 0.6.*cos(2*pi*1*fa_ground*t); % Frequency can be changed here


x2 = 0.5.*cos(2*pi*2*fa_ground*t); % Frequency can be changed here
x3 = 0.4.*cos(2*pi*3*fa_ground*t); % Frequency can be changed here
x4 = 0.3.*cos(2*pi*4*fa_ground*t); % Frequency can be changed here
x5 = 0.3.*cos(2*pi*5*fa_ground*t); % Frequency can be changed here
a = (x1+x2+x3+x4+x5); % Composed signal

x6 = 0.6.*cos(2*pi*fb_ground*t); % Frequency can be changed here


x7 = 0.5.*cos(2*pi*2*fb_ground*t); % Frequency can be changed here
x8 = 0.4.*cos(2*pi*3*fb_ground*t); % Frequency can be changed here
x9 = 0.3.*cos(2*pi*4*fb_ground*t); % Frequency can be changed here
x10 = 0.3.*cos(2*pi*5*fb_ground*t); % Frequency can be changed here
b = (x6+x7+x8+x9+x10); % Composed signal

Ampa=db2mag(65); % Set amplitude to 65 dB


Ampb=db2mag(75); % Set amplitude to 75 dB
dBa=mag2db(Ampa);
dBb=mag2db(Ampb);

a=Ampa.*a; % Change amplitude of signal a to 65 dB


b=Ampb.*b; % Change amplitude of signal b to 75 dB

tune=[a,b]; % Create the tune


sound(tune) % Play the tune in Matlab

figure(1)
plot(Fs*t(1:50), abs(20 * log10(a(1:50))), 'r', Fs*t(50:100), abs(20 *
log10(b(50:100))))
xlabel('time [miliseconds]')
ylabel('signal amplitude [dB]')
legend('Signal a', 'Signal b')
hold on
plot(Fs*t(1:50), dBa, 'r', Fs*t(50:100), dBb)
hold off

NFFT = 2^nextpow2(L); % next power of 2 from lenght of y


A=fft(a,NFFT)/L;
c=Fs/2*linspace(0,1,NFFT/2+1);

Page 12
B=fft(b,NFFT)/L;
d=Fs/2*linspace(0,1,NFFT/2+1);

figure(2)
plot(c,2*abs(A(1:NFFT/2+1)), d,2*abs(B(1:NFFT/2+1)) )
title('Amplitude spectrum of y(t)')
xlabel('Frequency [Hz]')
ylabel('Amplitude')
legend('Signal a', 'Signal b')

figure(3)
plot(Fs*t(1:25), x1(1:25), Fs*t(1:25), x2(1:25), Fs*t(1:25), x3(1:25),
Fs*t(1:25), x4(1:25), Fs*t(1:25), x5(1:25))
title('Signal a')
legend('Fundamental', '1^{st} harm.', '2^{nd} harm.', '3^{th} harm.', '4^{th}
harm.')
xlabel('time[ms]')
ylabel('signal amplitude')

figure(4)
plot(Fs*t(1:25), x6(1:25), Fs*t(1:25), x7(1:25), Fs*t(1:25), x8(1:25),
Fs*t(1:25), x9(1:25), Fs*t(1:25), x10(1:25))
title('Signal b')
legend('Fundamental', '1^{st} harm.', '2^{nd} harm.', '3^{th} harm.', '4^{th}
harm.')
xlabel('time[ms]')
ylabel('signal amplitude')

Page 13

Centres d'intérêt liés