a signal and sensor design
MECH4720: SENSORS & SIGNALS
ASSIGNMENT 01: SONAR AID FOR THE BLIND
Name: Mao SHAN
Student Number: 308038746
Email: msha3421@mail.usyd.edu.au
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 2
Contents
1 Brief Design Specifications .................................................................... 3
2 Sensor Array Design ............................................................................... 3 2.1 Detection of Obstacles Range and Angles ............................................................. 3 2.2 Sensor Selection ............................................................................................................ 4 2.3 Arrangement of Sensor Array .................................................................................... 6
3 Modulation Method and Signal Processing .......................................... 6
3.1 Determining maximum pulse width .......................................................................... 7 3.2 Determining Maximum Pulse Repetition Frequency ........................................... 8 3.3 Determine the Bandwidth of Matched Filter ........................................................... 8 3.4 Timing Process .............................................................................................................. 9 3.5 Improving Azimuth Accuracy in Front Area ........................................................... 9
4 Calculations ........................................................................................... 11 4.1Calculation of Angle and Range ............................................................................... 11 4.2 Minimum Detection Range ........................................................................................ 13 4.3 Maximum Detection Range and Required Amplifier Gain ................................ 13 4.4 Calculation of Time Difference ................................................................................. 14 4.5 Accuracy of Angle ....................................................................................................... 15 4.6 Accuracy of Range ...................................................................................................... 18
5 System Simulations .............................................................................. 19
6 Signal Processor Design ...................................................................... 21 6.1 System Block Diagram ............................................................................................... 21 6.2 Measurement of Range and Angle .......................................................................... 21 6.3 Audio and Tactile Output ........................................................................................... 22
6.3.1 Audio Output ............................................................................................................. 22 6.3.2 Tactile Output ............................................................................................................ 23
7 System Considerations ......................................................................... 23
7.1 Consideration of Power Supply ............................................................................... 23 7.2 Consideration of Mechanical Mounting ................................................................ 24 7.3 Consideration of Interference .................................................................................. 25
8 References ............................................................................................. 26
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 3
1 Brief Design Specifications z Wearable and unobtrusive – total mass < 1kg including batteries. z Detection range: up to 5m. z Range accuracy: 10cm z Instantaneous field of view in azimuth: +/-60°. z Azimuth resolution: <1°around the forward sector degrading to 5°at 60°. z Instantaneous field of view in elevation: 45°around the forward sector, can be worse at larger
angles. z Elevation resolution: good enough to identify broken/ irregular pavements and curbs as well as
overhanging branches and signs. z Multiple outputs – tactile and audio. z Operation for up to 8 hours without recharging.
2 Sensor Array Design As mentioned in the system specifications, the system should detect not only the distance between the obstacles and user, but also the azimuth and elevation angles of the obstacles. Hence, appropriate design of the sensor array becomes an important step in our system. 2.1 Detection of Obstacles Range and Angles Using Time-of-Flight technique, we know that it’s easy to determine the range of the obstacles. However, this technique can not be used to locate the angle of the obstacle at the same time. To detect the azimuth and elevation angles at the same time, another technique called “echolocation” is introduced into our system design. Actually, it’s a derived version of the Time-of-Flight technique. The right figure shows the principle of the two-dimensional echolocation. The sensor array shown in the figure is composed by one ultrasonic transmitter in the center, and two receivers on the left and right sides. Ultrasonic pulses are transmitted, then reflected by the obstacle, and finally received by two receivers. In the figure, θ is the azimuth angle of the obstacle, and DL and DR are the distances from obstacle to left and right receivers respectively, while R is the range from transmitter to obstacle. If we time the intervals ΔTL and ΔTR from transmitting a
θ DL DRR
Left Receiver Right ReceiverTransmitter
Obstacle
d d
Figure 2-1 Principle of two-dimensional echolocation
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 4
pulse until the echo reflected by the obstacle is received by the left receiver and the right receiver, we can get the DL and DR using time-of-flight technology hence the azimuth angle θ and range R can be worked out using some equations. Refer to Section 4 for detailed calculation. Using the same principle, two more receivers should be added on up and down sides of the transmitter to detect the elevation angle, which are illustrated as Figure 2-2. In the same way of calculating θ, the elevation angle φ can be got. Therefore, the combination of one transmitter and four receivers can realize function of tree-dimensional echolocation. Let’s call this combination a “Measurement Unit”. 2.2 Sensor Selection The specification of the instantaneous field of view is +/-60° in azimuth, and 45° in elevation. To reach this, those ultrasonic sensors with very wide beam width are required. I searched for many popular sensors on websites. Most of them only have beam width of 55°, 50°, 36° or even 10°, which are far from satisfying our requirements. So I summarized two options to this problem: 1. Mechanical scanning. 2. Multiple measurement units aiming different directions. The first one, mechanical scanning involves a motor driving the measurement unit to scan from left to right. I think it’s very power consuming, and with low efficiency. And more importantly, the mechanical scanning can not produce required instantaneous field of view as mentioned in the specification. My choice is the latter option, because they are static, consuming less power. Several measurement units aiming different directions can wider the instantaneous field of view. This issue will be discussed in 2.3. The ultrasonic transmitter and receiver selected are produced by Murata Manufacturing Co. Ltd. Part number of the transmitter and receiver are MA40B8S and MA40B8R respectively. MA40B8S and MA40B8R have the following features. z Compact and light weight. z High sensitivity and sound pressure. z Less power consumption. z High reliability. They have the same dimensions, see Figure 2-3. And Figure 2-4 is the directivities of the transmitter MA40B8S and the receiver MA40B8R.
R
R
R
TR
d
Left Right
Up
Down
d T
R
R R
R
Left Right
Up
Down
Figure 2-2 Sensor Array
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 5
(a)
(b)
Figure 2-3 Dimensions of Figure 2-4 Directivity of the transmitter MA40B8S (a) MA40B8S and MA40B8R and the receiver MA40B8R (b) Please refer to Table 2-1 for detailed information of MA40B8S and MA40B8R.
Table 2-1 Specification of MA40B8S and MA40B8R Global Part Number MA40B8S MA40B8R Previous Part Number Construction Open struct. Open struct. Using Method Transmitter Receiver Nominal Freq. 40kHz 40kHz Overall Sensitivity Sensitivity -63dB typ. (0dB=10V/Pa) S.P.L. 120dB typ. (0dB=0.02mPa) Directivity 50° (typ.) 50° (typ.) Cap. 2000pF 2000pF Min. Using Temp. Range -30°C -30°C Max. Using Temp. Range 80°C 80°C Min. Detectable Range 0.2m 0.2m Max. Detectable Range 6m 6m Resolution 9mm 9mm Max. Input Voltage 40Vp-p Continuous signal Note
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 6
2.3 Arrangement of Sensor Array Given that the typical beam widths of both transmitter and receiver we use are 50°, the specification of 45° instantaneous field of view in elevation is reached. However, only one transmitter can not fulfill the specification of the azimuth one, i.e. +/-60°. See Figure 2-5. Actually, 3 transmitters (corresponding to 3 measurement units) aiming different directions are needed to cover the required instantaneous field of view in azimuth. In order to get a total azimuth beam width equal or larger than 120° (+/-60°), the three measurement units should be arranged with 40° angle difference. See Figure 2-6 for the arrangement of three measurement units. And Figure 2-7 shows the combined azimuth beam width of three measurement units.
3 Modulation Method and Signal Processing Time-of-flight technique usually uses pulsed signal, or pulse amplitude modulation. In this type of modulation, we need to determine two parameters, maximum pulse width and maximum pulse repetition frequency, according to the specification of range resolution and maximum detection range. When processing echo signals from obstacle, the appropriate bandwidth of matched filter should be determined. To extract range information from the received echoes, timing process is designed in this section. And to calculate the angle of the obstacle, we still need to detect the time difference of the two receivers receiving echoes from the obstacle.
Azimuth Beam Width of One Measurement Unit
25° 25°60 ° 60°
Figure 2-5 Azimuth Beam Width of One Measurement Unit
Unit 2
Unit 1 Unit 3
Unit 1
Unit 2
Unit 3
40° 40°
Figure 2-6 Arrangement of Three Measurement Units
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 7
3.1 Determining maximum pulse width The pulse width τ determines the ability of resolving two closely spaced obstacles. If the two obstacles are too close, only a single echo would be received. In Figure 3-1, which is a time-of-flight system, the width of transmitted pulse isτand the distance between two obstacles is Δd. Figure 3-1 Time-of-flight system From Figure 3-1, we can clearly see that, to avoid overlapping of two reflected echoes from two closely spaced obstacles, theτmust be smaller than a certain value, that is,
v dΔ
< 2
τ (3-1)
where v is the propagation velocity of sound in air. In our system, the specification of the range resolution is 10cm, so according to Eq. (3-1), the maximum pulse widthτ= 0.2 / 340 (m/s) = 588.2 μs. We will use 550μs as the width of pulse transmitted in our system, because 550 μs is 22 wavelengths of 40 kHz ultrasonic wave.
Azimuth Beam Width of Measurement Unit 2
40° 40°6 0° 60°
Azimuth Beam Width of Measurement Unit 1
Azimuth Beam Width of Measurement Unit 3
Figure 2-7 Azimuth Beam Width of Three Measurement Units
Transmitter
Receiver
Obstacle 1 Obstacle 2
Transmitted Pulse
Echo from Obstacle 1
Echo from Obstacle 2
Δd
τ
τ
τ
Figure 3-1 Time-of-Flight System
Transmitter
Receiver
Obstacle
The Second Transmitted Pulse
Echo of The First Pulse
Figure 3-2 Pulse Repetion Freqency Too High
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 8
3.2 Determining Maximum Pulse Repetition Frequency If the pulse repetition frequency is too high, the second pulse is transmitted before the echo of the first pulse is received, which is illustrated in Figure 3-2, the receiver may mistake the first echo as the return signal of the second transmitted pulse, thus calculate a wrong range value. For this reason, we should determine an appropriate pulse repetition frequency in our system to avoid this mistake happening. The maximum repetition frequency fmax can be acquired from the following equation,
max max 2R
v f = (3-2)
where v is the propagation velocity of sound in air Rmax is the maximum detection range.
The maximum detection range required in our system is 5m, but in my opinion, the receiver may still have the ability to receive echoes from obstacles further than 5m, so the Rmax should not be 5m but 10m, in order to make sure that the system will never receive the echo of the first pulse after transmitting the second pulse. Hence, using Eq. (3-2), we can get the maximum pulse repetition frequency in our system. fmax = 340/ (2 * 10) = 17 Hz. But in our system, I think 10 Hz is enough. 3.3 Determine the Bandwidth of Matched Filter Matched filter is used in our system to reduce the disturbing of noise. The center frequency of matched filter is equal to the signal transmitted, i.e. 40 kHz. As a rule of thumb, the suitable bandwidth β of the matched filter is determined by τ, which is the width of transmitted pulse.
τ β
1 = (3-3)
In Section 3.1, the pulse width is 550 μs. Substituting theτ= 550 μs into Eq. (3-3), we get β = 1 / 550 μs = 1.82 kHz In addition, we have to consider the Doppler Effect at the same time. Because maybe the pedestrian is walking or there is a moving obstacle. Suppose that the maximum relative velocity between the pedestrian and the obstacle is -10m/s (obstacle is approaching). According to Doppler equation
sr fvc vc
f + −
= (3-4)
where c is the propagation velocity of sound in air v is the relative velocity between pedestrian and the obstacle Substituting v = 10m/s, c = 340m/s and fs = 40kHz into the Eq. (3-4), we get fr = (340+10) / (340-10) * 40000 = 42.42 kHz So the Doppler shift frequency is fd = fr – fs = 42.42k – 40k = 2.42 kHz
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 9
This Doppler shift frequency should be within the bandwidth of the matched filter. So the Eq. (3-3) should be rewrote as
df2 1 +=
τ β (3-5)
And the appropriate bandwidth of the matched filter should be β = 1.82 kHz + 2 * 2.42 kHz = 6.66 kHz 3.4 Timing Process To determine the range from the obstacle to the receiver, the simplest way is to time the interval from transmitting the pulse until receiving the echo signal. See Figure 3-3 for the principle of timing process used in TOF.
Bang Pulse
Noise Level
Echo from Obstacle
t ΔT
Detection Threshold
Figure 3-3 Principle of timing process used in TOF
For a measurement unit in our system, there are four receivers, so we need four individual timers to do this job. Every timer measures the time used from pulse transmitting and receiving. Thus, we can get DL, DR, DU and DD at the same time for a measurement unit. Theoretically, the range accuracy only depends on the accuracy of the timer, the higher accuracy the timer have, the more accurate range we get. In actual practice, the timing process usually doesn’t have very high accuracy because of timing errors, such as noise, variation in sound speed under different temperature, and “Range Walk”. 3.5 Improving Azimuth Accuracy in Front Area In echolocation technique, it’s very important to know the difference between the distances from the obstacle to left receiver and right receiver, to calculate the angle of the obstacle. The distance difference can be calculated if we know the time difference between the left receiver and right receiver receiving the echoes from the obstacle. The time difference could be very small, because a little bit change of obstacle angle leads to very small change of distance difference. Theoretically, it’s not a problem as long as we have enough accuracy to measure the distance difference. However, using TOF technique in practical application, the distances measured from receivers to the obstacle may have about several millimeters error due to timing errors, which will seriously affect the accuracy of the angle calculated. In our system, the specification of azimuth accuracy in front area should be less than 1°, and less than
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 10
5° on the sides. In Figure 2-6, the measurement unit 2 is aiming the front area so the azimuth accuracy of this measurement needs special improvement, while the unit 1 and 3 do not need improvement because the accuracy requirement is lower. Hence, in order to increase the accuracy of time difference of echoes received by two receivers of measurement unit 1, I measure phase difference of two echoes in addition to measuring the time difference by timers, instead of the way only by timers. So the time difference is composed by two parts, raw time measured by timers and more accurate time calculated by phase difference. In our application, I add a phase detector to measure the phase difference ΔΦ. As shown in Figure 3-4, suppose two received echo signals with phase difference of ΔΦ, which has range from –π toπ. The phase detector I use has proportional voltage output according to the phase difference ΔΦ. The equation of output voltage V and phase ΔΦ is
π/5 ΔΦ×=V (3-6)
ΔΦ
t
t
Amplitude
Figure 3-4 Two echo signals with phase difference of ΔΦ
V
Φπ
5
-π
-5
Figure 3-5 Phase Detector Function
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 11
4 Calculations In this section, some equations are introduced to calculate the angle and range of the obstacle. Besides, the minimum and maximum detection range will be calculated as well. And performances of the timers and phase difference detector are considered to meet the accuracy specification of angle. 4.1Calculation of Angle and Range Now let’s come back to the theory of echolocation. From Figure 4-1, which is the mathematical model of two-dimension echolocation, we have
θθθ sin2cos)sin( 2222 dRdRRdRDL ++=++= (4-1)
θθθ sin2cos)sin( 2222 dRdRRdRDR −+=+−= (4-2)
Where R is the range from obstacle to transmitter d is the distance between transmitter and receiver θ is the azimuth angle of the obstacle So
2 2 222 dDD
R RL −+
= (4-3)
⎟ ⎟ ⎠
⎞ ⎜ ⎜ ⎝
⎛ − =
dR DD RL
4 arcsin
22
θ (4-4)
From Figure 2-3, we can get the diameter of the transmitter and receiver sensor, i.e. 0.016m. So I think the appropriate distance between transmitter and receiver, which is d, could be 0.02m. We can use ΔD to represent DL – DR, to view the relationship of DL, ΔD and R when d is 0.04m, see Figure 4-2. And Figure 4-3 is the relationship of DL, ΔD and θ. Using the same style of Eq. (4-4), the elevation angle φ can be calculated through the following equations
⎟ ⎟ ⎠
⎞ ⎜ ⎜ ⎝
⎛ − =
dR DD UD
4 arcsin
22
ϕ (4-5)
where DD and DU are the distances from obstacle to down and up receivers respectively. R is the range from obstacle to transmitter, which has been calculated using Eq. (4-3). Now it’s clear that, if we can measure the DL, DR, DU, DD, we can calculate range R, azimuth angle θ and elevation angle φ using the Eq. (4-3), (4-4) and (4-5).
θ DL DRR
RL: Point of Left Receiver RR: Point of Right Receiver T: Point of Transmitter
Obstacle
RL(-d, 0) T(0, 0) RR(d, 0)
A(R*sinθ, R*cosθ)
X
Y
A: Point of Obstacle
Figure 4-1 Mathematical model of two-dimension echolocation
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 12
Figure 4-2 Relationship of DL, ΔD and R with color encode by R
Figure 4-3 Relationship of DL, ΔD and θ with color encoded by θ
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 13
4.2 Minimum Detection Range When the transmitter sends a pulse, the receiver will immediately receive a “bang” pulse with large amplitude, which leaks from the transmitter. If an obstacle is located too near the transmitter and receiver, the echo from the obstacle and the bang pulse will overlap so that the receiver can not identify the echo from the obstacle. See Figure 4-4 (a) for normal situation of bang pulse and echo, and (b) for overlapping bang pulse and echo. The minimum detection range Rmin has something to do with the transmitted pulse widthτ.
2min v
R τ
= (4-6)
where v is the propagation velocity of sound in air In Section 3.1, τis calculated with maximum value of 588.2 μs. So the Rmin = 588.2 * 10
-6 * 340 / 2 = 0.1m. 4.3 Maximum Detection Range and Required Amplifier Gain The maximum detection range depends not only on the power of pulse transmitted and sensitivity of receiver, but also on miscellaneous factors, such as noise, modulation and demodulation methods and so on. In this section, only power of pulse transmitted and receiver sensitivity are considered in calculation of maximum detection range. From Table 2-1 in Section 2.2, we can get the sound pressure of the transmitter MA40B8S and sensitivity of the receiver MA40B8R we selected in our system. The typical SPL of MA40B8S is 120dB (0dB=0.02mPa), 0.02mPa = 20μPa, which is the standard reference level for SPL. The voltage applied in the transmitter is 10V, which is typical value. So the sound pressure level at 30cm range is
dBLp 120=
The maximum detection range required of our system is 5m, so the beam-spread loss at 5m if given by
dB R R
H ref
4.24 3.0
5 log20log20 1010 ===
The Bang Pulse
Echo from Obstacle t
The Bang Pulse
Echo from Obstacle t
Figure 4-4 (a) Bang pulse and echo under normal situation (b) Overlapping bang pulse and echo
(a)
(b)
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 14
Suppose that our system can detect an obstacle with target strength T=-10dB at 5m range. Hence the SPL back at the receiver is
dBTHLL pr 2.61104.2421202 =−×−=+−=
So we have received pressure
11481010 06.320 === rL
P (Relative to 0.0002μbar) Therefore
barP μ230.00002.01148 =×=
The typical sensitivity of the receiver is -63dB (0dB=10V/Pa), 10V/Pa = 1V/μbar, thus the sensitivity
response is barV μ/1008.710 420 63
− −
×=
So the output voltage of receiver should be
VVout μ1631008.7230.0 4 =××= −
Suppose the processing circuit after receiver and amplifier can accept signal with amplitude over 100mV, the gain of the amplifier circuit required is
dB V V
G out
req 8.55 1063.1 1.0
log20log20 41010 =
× ==
−
Therefore, to detect the echo signal from an obstacle with target strength -10dB at 5m range, the amplifier after the receiver should have gain of at least 55.8dB. 4.4 Calculation of Time Difference For measurement unit 1 and 3, they don’t need to satisfy high angle accuracy because they are aiming side directions. And the required elevation accuracy of all measurement units is not high either. So the accuracy of time measured by timers is enough. So the time difference is calculated by following equation.
RL TT −=ΔT (4-7)
For the left and receiver of measurement unit 2, which is aiming the front area, we add a phase detector to get a more accurate time, thus to improve the azimuth accuracy. Knowing the raw time measured by timers and phase difference of two channels of echoes, we can calculate the time difference ΔT using the following equation
T T
T T
T T RL
π φ
22 ]
2 []
2 [T
Δ +
⎭ ⎬ ⎫
⎩ ⎨ ⎧
−=Δ (4-8)
where TL and TR are the time measured by left and right receivers respectively ΔΦ is the phase difference of echoes received by two receivers
T is the cycle time of the pulse carrier signal The “[]” signs in Eq. (4-8) mean to get the largest integer less than the value in “[]”.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 15
In our system, the cycle time of the pulse carrier signal is 25μs. To use Eq. (4-8), the timers must have accuracy of at least T/2, i.e. 12μs. 4.5 Accuracy of Angle From Figure 4-3, the relationship of DL, ΔD and θ can be clearly seen. However, it is still not exact enough to determine the accuracy of angle. Let’s plot two 2D figures, Figure 4-4 and Figure 4-5, with DL = 2.5m and θ = 20° respectively.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
26
28
30
32
34
36
38 Relationship of DL and Theta with Delta Distance = -0.02 m
DL (m)
T he
ta (
de gr
ee )
Figure 4-4 Relationship of DL and θ when ΔD = -0.02m. In Figure 4-4, when DL is getting smaller, the θ becomes more sensitive to the change of DL. We can conclude that, the smaller DL is, the lower accuracy of angle becomes at a given ΔD. That’s to say, the angle accuracy is the lowest when DL reaches its minimum detectable value. In Section 4.2, the minimum detection range has been calculated, which is 0.1m. So try to satisfy the specification of the azimuth angle resolution, we should analyze the accuracy of angle when DL reaches minimum value, i.e. 0.1m. The relationship of ΔD and θ when DL = 0.2m is shown as Figure 4-5.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 16
-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 -65
-60
-55
-50
-45
-40
-35
-30
-25
-20
-15
X: 0.0135 Y: -60.05
Relationship of Delta Distance and Theta with DL = 0.1 m
DL-DR (m)
T he
ta (
de gr
ee )
X: 0.0102 Y: -55.04
-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 -25
-20
-15
-10
-5
0
5
10
15
20
25
X: 0 Y: 0
X: 0.0006 Y: -0.8771
Relationship of Delta Distance and Theta with DL = 0.1 m
DL-DR (m)
T he
ta (
de gr
ee )
(a) (b)
Figure 4-5 Relationship of ΔD and θ when DL = 0.1m.
(a) measurement unit 1 (b) measurement unit 2 (c) measurement unit 3
(c) The specification of azimuth angle resolution in front area is less than 1° and less than 5° at 60°. From the data points in Figure 4-6 (a) and (c), to achieve accuracy of 5° at -60° and 60°, the system should have the ability to identify the ΔD of 3.3mm. So the time difference is 3.3*10-3 / 340 = 9.7μs. Hence every timer should have accuracy of at least 4.8μs. To satisfy accuracy of 1° at front area, the ΔD is 0.6mm, which can be seen in Figure 4-6 (b), so the time difference is 6*10-4 / 340 = 1.8μs So the timer used in our system should have accuracy of at least 0.9μs. If the system is ideal, it’s not a problem. As stated in Section 3.4 and 3.5, the time difference measured actually may satisfy the accuracy of 9.7μs, but can not achieve accuracy of 1.8μs. So a phase detector is used to improve only the azimuth accuracy at front area. Using Matlab, we can know the phase difference when the azimuth angle of the obstacle is 1° and range is 0.1m, see Figure 4-6. From the Figure 4-6, it can be seen that the ΔΦ is 29°, and the output voltage of the phase detector is 1.61V calculated by Eq. (3-6). Suppose that the AD converter our system used has the resolution of 8bit, i.e. 39mV with input range of -5V to 5V. It’s totally enough to identify the ΔΦ when the azimuth angle of the obstacle is 1°.
-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 15
20
25
30
35
40
45
50
55
60
65
X: -0.0134 Y: 60
Relationship of Delta Distance and Theta with DL = 0.1 m
DL-DR (m)
T he
ta (
de gr
ee )
X: -0.0101 Y: 54.95
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 17
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 -1
-0.5
0
0.5
1 Echo Signals, with Delta Phase = 28.9924 degree, Phase Detector Output = 1.6107 V
Le ft
R ec
ei ve
r S
ig na
l ( V
)
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 -1
-0.5
0
0.5
1
R ig
ht R
ec ei
ve r
S ig
na l (
V )
Time (ms)
Figure 4-6 Phase difference when azimuth angle of obstacle is 1° at range of 0.1m The accuracy of elevation angle is equal to the azimuth accuracy of measurement unit 1 and 3, because they have the same principle of measurement. The accuracy of timer selected is 4.8μs or ±2.4μs, so the clock frequency of the timers is at least 416 kHz. However, I prefer using 1MHz clock frequency of timers, because it’s better that the system can acquire higher accuracy. Now we can summarize the accuracy of angle in our system. z The accuracy of azimuth angle at front area is less than 1°. z The accuracy of azimuth angle at sides is 5°. z The accuracy of elevation angle is 5°.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 18
4.6 Accuracy of Range First we should make sure the relationship of DL and R with a given ΔD.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5 Relationship of DL and Range with Delta Distance = -0.02 m
DL (m)
R an
ge (
m )
Figure 4-7 Relationship of DL and R when ΔD = -0.02m. From Figure 4-7, we can see that, the R is proportional to DL. So we do not need to find out a point that has the worst accuracy. Let’s plot a figure illustrating relationship of ΔD and R with a given DL.
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 2.48
2.485
2.49
2.495
2.5
2.505
2.51
2.515
2.52
2.525
X: 0 Y : 2.5
Relationship of Delta Distance and Range with DL = 2.5 m
Delta Distance (m)
R an
ge (
m ) X: 0.0033
Y : 2.502
Figure 4-8 Relationship of ΔD and R when DL = 2.5m. As calculated in Section 4.4, the accuracy of ΔD in our system is 3.3mm. It can be seen from Figure 4-8 that the accuracy of range is 2mm.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 19
5 System Simulations Matlab is used to simulate the system. The following are some parameters of our system related to the simulation. Frequency of transmitted pulse signal fs: 40 kHz Width of transmitted pulseτ: 550 μs Frequency of pulse repetition fp: 10 Hz Distance between transmitter and receiver d: 0.04 m Now we assume an environment for this simulated system. The actual range of an obstacle Ra: 2.8 m The actual azimuth angle of the obstacle Aa: 15° And both the pedestrian and the obstacle are static. Figure 5-1 shows the pulses transmitted and echoes received.
0 20 40 60 80 100 120 140 160 180 200 -1
0
1 Transmit Signal and Echo Signals
0 20 40 60 80 100 120 140 160 180 200 -1
0
1
0 20 40 60 80 100 120 140 160 180 200 -1
0
1
Time (ms)
Figure 5-1 Transmitted signals and echo signals Magnify the part of echo signals, we have the following figure.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 20
16.4 16.5 16.6 16.7 16.8 16.9 17 -1
-0.5
0
0.5
1 Echo Signals, with Delta Phase = 78.4595 degree, Phase Detector Output = 4.3589 V
Le ft
R ec
ei ve
r S
ig na
l ( V
) X: 16.48
Y: 0
16.4 16.5 16.6 16.7 16.8 16.9 17 -1
-0.5
0
0.5
1
R ig
ht R
ec ei
ve r
S ig
na l (
V )
Time (ms)
X: 16.45 Y : 0
Figure 5-2 Echo signals and output of phase detector From Figure 5-2, we can know the intervals of pulse transmission and reception read from left and right receiver timer are 16.48ms and 16.45ms respectively. And the phase difference of two echoes is 78.45°. Substituting these values to Eq. (3-7), we get
sμ π π
448.30448.55.12)13161318(25 2
180/45.78 2 25
] 25 164502
[] 25 164802
[T =+×−=× ×
+ ⎭ ⎬ ⎫
⎩ ⎨ ⎧ ×
− ×
=Δ
And the DL is calculated as follows
m vT
D LL 8016.22 34001648.0
2 =
× ==
So the DR can also be calculated as
mTvDD LR 79125.234010448.308016.2 6 =××−=Δ−= −
Substituting DL and DR into Eqation (4-3) and (4-4), we have
m dDD
R RL 7989.2 2
02.0279125.28016.2 2
2 222222 =
×−+ =
−+ =
°=⎟⎟ ⎠
⎞ ⎜⎜ ⎝
⎛ ×× −
=⎟ ⎟ ⎠
⎞ ⎜ ⎜ ⎝
⎛ − = 98.14
7989.202.04 79125.28016.2
arcsin 4
arcsin 2222
dR DD RLθ
We know that the actual range of the obstacle is 2.8m, hence our calculation of range only have error of 1.1mm. And the calculated azimuth angle of the obstacle only has error of 0.02° comparing to actual azimuth angle 1°.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 21
6 Signal Processor Design 6.1 System Block Diagram
Amp & Filter
Amp & Filter
Transmitter Driver
Timer
Timer
Phase Detector
Signal Processor
Measurement Unit 2
Measurement Unit 1
Measurement Unit 3
Audio and Tactile OutputAmp & Filter Timer
Amp & Filter Timer Up
Down
Left
Right
Figure 6-1 System Function Diagram As can be seen in Figure 6-1, the system is composed by three measurement units, audio and tactile output and miscellaneous modules, such as power supply, and AD converter, which are not illustrated in this diagram. For arrangement of these three measurement units, see Figure 2-6 in Section 2.3. The measurement unit 2 consists of one transmitter and four receivers. Each receiver is connected to an amplifier and then a timer. There is a phase detector connecting to the left and right receiver, because of the high requirements of the azimuth accuracy around the forward sector, for detailed information, see Section 3.5. The measurement unit 1 and unit 3 are the same to the unit 2 except that they do not have phase detectors, because there’s no high requirement of azimuth accuracy except forward sector. The range and angle information measured is then converted to audio and tactile signals and sent to audio and tactile output module. See Section 6.3. 6.2 Measurement of Range and Angle All of three transmitters will transmit a pulse simultaneously. The timing is started at the same time, and then system is waiting for echo signals. Echo signals are received by receivers and then amplified and filtered. They will trigger the threshold comparators to stop timing. Measured time information of all timers is read by processor, which will calculate the range and angles of the obstacle using the equations we’ve discussed. In addition, the echoes received by left and right receivers of measurement unit 2 are passed to the phase detector. Voltage output of the phase detector is AD converted to digital value and then read by process to analyze. The signal process flow is shown as Figure 6-2.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 22
Amplif ied & Filtered
Stop TimingStart Timing
Transmit Pulses
Wating for Echoes
Calculate Range & Angle
Phase Detector AD Convert
Only f or Lef t and Right receiver of Measurement Unit 2
Figure 6-2 Signal process flow in measuring 6.3 Audio and Tactile Output The obstacle information measured, such as range and angle, should be ultimately converted to some types of signal that the blind people can recognize. This process is to remind the blind people that something is in front of him or her, how far it’s away, and what’s the direction of the obstacle. Usually audio, tactile, or both of them are used in typical sonar aid system. The specification of our system requires multiple outputs, i.e. both tactile and audio. 6.3.1 Audio Output The range and angle information has been acquired by the processes described previously. The first thing to convert is the range information of obstacle ranging from 0.1m to 5m. And both the azimuth and elevation angles are to be converted as well. Headphones are used to in the audio output. The range information is represented by the overall amplitude of the beep sound. When the obstacle is far away, the overall amplitude of the beep sound is low just reminding that there’s an obstacle ahead, while there’s ah obstacle very near, the audio output module will produce louder beep sound to warn the user that the obstacle is very near. To represent the azimuth angle of the obstacle, I use the spatialization principle [2], to produce stereo sound as if placing the sound source to a given point of the space related to the listener’s head, see Figure 6-3. The listener will hear sounds with intensity and phase binaural differences. So this process is to transform the mono beep sound to binaural signals with different amplitudes, according to the value of azimuth angle. And the elevation angle is transformed to the frequency of the beep sound in our system. The frequency is proportional to the elevation angle. The higher the obstacle is, the higher frequency beep sounds produced. Hence the relationship between the beep sounds produced and range, azimuth angle and elevation angle.
Left Ear Right Ear
Sound Source
θ
Figure 6-3 Spacialization Principle
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 23
z Elevation angle controls the frequency of the beep sounds z Range controls the overall amplitude of the beep sounds z Azimuth angle controls the amplitudes of both left and right beep sound channels separately The whole audio output module has the following function diagram.
Beep Sound
Generator
Elevation Angle
Overall Amplifier
Left Channel Amplifier
Right Channel Amplifier
Range
Azimuth Angle
Output
Output
Control Frequency Control Amplitude Control Amplitude
Control Amplitude
Figure 6-4 Audio Output Module Function Diagram
6.3.2 Tactile Output Tactile output is a complementary output in our system. The tactile output usually uses small motors to generate vibration that can be known by user, just like vibration mode of a mobile phone. However, motors even small ones can consume a lot of energy, so we only use as less motors as possible to save energy and prolong the operation time of our system. The tactile output module plays a complementary role to the audio output, because the tactile output can not present as much information as the audio output. However, it may become the only information source the user can rely on when there’s very loud noise in the environment to disturb the user to hear the audio output. The tactile output in our system only need to output the range information, which is the most important information indicating how far away the obstacle is. And the vibration frequency is proportional to the range value.
7 System Considerations 7.1 Consideration of Power Supply The power supply chose for our system is Lithium-ion battery. It’s very suitable for portable electronic equipments with one of the best energy-to-weight ratios, no memory effect, and a slow loss of charge when not in use. Of course we need to design a recharge circuit in our system. To calculate the operation time of this system, we have to evaluate the power consumption first. Transmitters and receivers are very power saving, because the transmitters only transmit 550μs width pulses with 10Hz repetition frequency, and the receivers are passive sensors without consuming
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 24
energy at all. There are some parts of our system consuming energy a lot. z Amplifiers z Processor z Audio output z Tactile output Every amplifier in our system is composed by some operational amplifiers to get required gain. I think each amplifier consumes 25mW, There are 16 amplifiers in total in our system, so they consume 400mW totally. The processor approximately consumes 50mW. The audio output should drive two speakers in headphones. Every speaker consumes 20mW, so they consume 40mW totally. The tactile output only has one small motor, but it will consume approximately 100mW. Energy consumed by other parts of the system may have 60mW in total. Hence, the total power needed for our system is 400+40+50+100+60 = 650mW. The voltage used in our system is 5V, so the current is 130mA. However, the nominal voltage of the Lithium-ion battery is 3.7V, so the current from Lithium-ion battery is 130 * 5 / 3.7 = 176mA. The required operation time of this system is 8 hour. Hence the minimum capacity required is 176 * 8 = 1408 mAh. Actually, the capacity of Lithium-ion battery selected should be larger than the value we’ve calculated, due to the energy loss in voltage conversion and some other reasons. So finally the capacity of the Lithium-ion battery selected is 1600 mAh. 7.2 Consideration of Mechanical Mounting As illustrated in Figure 2-6 in Section 2.3, this system is light and portable. It’s suitable for mounting on head or hat, and can also be used along with a white cane. The Figure 7-1 shows the different mounting positions of this system.
Headphone
Sonar Aid System
Tactile Output
Headphone
Sonar Aid System
Tactile Output
(a) (b)
Figure 7-1 Sonar Aid System mounting (a) on head or hat and (b) on a white cane
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 25
7.3 Consideration of Interference If there are several blind people using the same sonar aid systems in the same area, the normal operations of systems will be disturbed by each other. Suppose there are two blind people using two same sonar aid systems A and B at the same place, when they transmit two pulses separately, A and B all receive the echoes reflected by the obstacle. But A and B do not know which echo is the return signal of its own. A or B may simply consider the one that first arrived as the echo of the pulse it transmitted, which may lead to wrong measurement result. That’s an example of interference existing when there are two systems. The situation is more complex when there are more systems in the same area. To reduce the happening chance of this interference as much as possible, I think there are several improvements and suggestions. z Reduce the pulse repetition frequency. Since the maximum range is 5m, the time used in each
cycle of measurement is about 30ms.Reducing the pulse repetition frequency can reduce the chance that the operation cycle of two system overlap, thus reduce the chance of interference.
z Get average data after multiple measurements. We can measure the range and angle several times, filter the bad data, and calculate the range and angle using average data. Thus reduce the happening of wrong calculation result when there’s interference.
z Think about some modulation methods that can add identification information into the pulses transmitted. For example, every system will generate a random small frequency shift value and add this shift value to the carrier frequency of the pulse when transmitting, so that the system has the ability to identify which echo belongs to it, avoiding mistaking.
The first and second methods mentioned can be realized in our system, because they don’t involve any change of hardware, but only processing algorithm and software. But the third one needs a center frequency selectable matched filter and more complex processing circuit, so it’s only a good suggestion for the further development of this system.
Assignment 01: Sonar Aid for the Blind
Mao SHAN 308038746 MECH 4720 Sensors and Signals 26
8 References [1] Kim, Yong-Jop., “Design Of Auditory Guidance System For The Blind with signal Transformation From Stereo Ultrasonic To Binaural Sound”, Department of Electrical Engineering & Computer Science, Korea Advanced Institute of Science and Technology, 2001. [2] Tatai, Peter., “Supterbat – A Navigation Aid For The Blind”, Department of Telecommunications and Telematics, Budapest University of Technology and Economics, Budapest, 2002. [3] Blenkhorn, P., Pettit. S and Evans. D.G., “An Ultrasonic Mobility Device with Minimal Audio Feedback:”, Department of Computation, UMIST, United Kingdom, 1997. [4] Brooker, G., MECH4720 Notes, Sensors & Signals, The University of Sydney, 2008 [5] “Electronic Travel Aids for the Blind”, http://www.noogenesis.com/eta/current.html