sonar characteristics
TRANSCRIPT
CHARASTERISTICS OF SONAR RANGE SENSOR SRF05
.A. Wickramasooriya, G. Hamilan, S.I.L. Jayawardena, L, W.M.D.L.W. Wijemanne, S.R. Munasinghe
Department of Electronics and Telecommunication Engineering
University of Moratuwa 10400, Sri Lanka
ABSTRACT
Sonar sensors are widely used in obstacle detection in
mobile robotic applications and certain limitations of these
sensors usually become the reason for failures in these
applications. Therefore, a thorough analysis of the actual
behavior of sonar sensors is required. This research studied
Devantech SRF05 sonar sensor, which is the most common
sonar sensor in practice. We have discovered its practical
beam pattern in 3-dimension (3D), optimum trigger
frequency, and maximum angle of detection.
Index Terms — Beam pattern, SRF05, sonar sensor, trigger
frequency, filtering algorithm.
1. INTRODUCTION
In mobile robotic applications sonar sensors are widely used
in obstacle detection and map building. However without a
proper knowledge of the sensor characteristics, and
limitations, it is difficult to achieve desired performance.
Here we present those limitations and techniques to
overcome them. We have used Devantech SRF05 sonar
sensor, which is widely used mainly due to range and low
cost. The main problem of sonar sensors is the lack of
knowledge of the 3D beam pattern of the sensor. The
provided data by the manufacturer [2] is only very general.
The trigger frequency of the sensor is also critical. We also
found that surface texture, audibility angles of the surface,
area of surface, and location of acoustic features on a
surface are also important facts for obstacle detection using
sonar sensors.
2. ANALYSIS OF THE BEAM PATTERN OF
DEVANTECH SRF05
(a)
(b)
Fig.2 (a) SRF05 sonar sensor, (b) Beam Pattern
Figure 2 shows SRF05 sonar sensor and its beam pattern.
We analyzed the behavior of the single sensor in a vacant
room, where echo was only received from a specific object.
We used a 25 × 25 cm plastic plate as the standard object
and gathered readings for different orientations. Figure 2
shows the experimental arrangement in that the object was
moved along horizontal and vertical straight lines in x-y
plane at several distances in z-axis.
Fig.2 Experimental Arrangement
x
z
y
h
z
Object
2.1 Accurate Beam Pattern
We approached in different ways to check the detectable
area of the sensor. First we measured the horizontal and
vertical width of the sensor beam at particular distances as
shown in Fig. 3
Fig. 3. Test to discover the beam pattern of SRF05 sensor
Fig. 4a and Fig. 4b shows the readings taken at distances of
d=z=100cm and d=z=300m. The data were collected from
d=z=0cm to d=z=400cm meters at 25cm separations. From
that we were able to identify two separate areas according to
the probability of detection.
Fig. 4a. Test data at the line {x, y=0, z=100}
Fig. 4b. Test data at the line {x, y=0, z=300}
Based on these results, we identified a high probability area
of detection (HPAD) and a low probability area of detection
(LPAD) using 0.7 probability of detection on the straight
line of {x, y=0, z=100} as shown in Fig. 5
Fig.2.4c. Area Seperation According to the Probability of
Detection at the line- {x, y=0, z=100}
Then we moved the object in z direction slowly (i.e. parallel
to the beam axis) and measure the readings and crossing
points of the beam. From that we could draw the beam
pattern of the sonar sensor for the detection of 25x25cm
object facing direct to the sensor.
We have placed the sensor horizontally, vertically and 45
degrees inclined as in Fig.2.5. The standard object was
moved in z-direction away from the sensor. We have taken
reading three times from the sensor with five sensors. It was
found that there were no major differences in the beam
pattern.
From the observation we could justify the above said
separation of the beam pattern into two different areas
depending on the probability of receiving an echo. From the
averaged value we have drawn the beam pattern.
Fig.2.5. Accurate Beam pattern of SRF05 sonar
Graphs included in Fig 2.6a, 2.6b and 2.6c below visualize
the test data received in this testing.
LPAD HPADD
x z
y
Fig.2.6a. Object moving along the line- {z, y=0, x=-10}
Fig.2.6b. Object moving along the line- {z, x=0, y=40}
Fig.2.6c. Object moving along the line-
{z, x=60×cos60, y=60×cos60}
Fig.2.7. Data Fluctuations
2.1.1 Accurate Beam Pattern - Results
The resulted 3-D beam pattern is illustrated below in several
2-D planes for the ease of the understanding. The planes are
named according to the axial representation shown in
Fig.2.2.
Fig.2.9. Beam pattern along the yz plane at x=-20
Fig.2.10. Beam pattern along the yz plane at x=-10
Fig.2.13. Beam pattern along the yz plane at x=10
Fig.2.12. Beam pattern along the yz plane at x=20
Fig.2.15. Beam pattern along the xz plane at y=-20
Fig.2.16. Beam pattern along the xz plane at y=-10
Fig.2.19. Beam pattern along the xz plane at y=10
Fig.2.18. Beam pattern along the xz plane at y=20
As described previously the area where we keep an object
and the probability of receiving echo is more than 0.7 is
defined as the high probability area of detection (HPAD).
This is covered along the beam axis. The area where object
is placed and the probability of receiving echo is less than
0.7 is defined as the low probability area of detection
(LPAD). This also depends on the size of the object. We
defined these areas according to our standard size object.
When the size of the object is even larger the HPAD may
expand into the LPAD.
Test 1 Results
Fig.2.20a and 2.20b illustrates data the variation of distance
when the standard object was moved at different distances
from the sensor along x – direction.
Fig.2.20a. Reading variation at along the line {x, y=0, z=368}
Fig.2.20b. Reading variation along the line {x, y=0, z=151}
2.3 Triggering Frequency
The SRF05 sensor only detects the first echo coming from
the object. We supply a short 10µs pulse to the trigger input
to start the ranging. The SRF05 will send out an 8 cycle
burst of ultrasound at 40 kHz and raise its echo line high. It
then listens for an echo, and as soon as it detects one it
lowers the echo line again. The echo line is therefore a pulse
whose width is proportional to the distance to the object. By
timing the pulse it is possible to calculate the distance. If
nothing is detected then the SRF05 will lower its echo line
after about 30ms.
The area power density (W/m2) of the echo declines as it
travels. We needed to set the triggering time of sensor such
that the power of the previous burst has declined such that it
will not affect the next trigger.
Fig.2.21. Effect of Multiple Echoes – Object at the point-
{x=0, y=0, z=200}
2.3.1 False Echoes
Multiple echoes are the result of an environment with many
objects. The sonar burst from sensor is spread to the
environment and may generate echoes from several objects
around the sensor. But the SRF05 is designed only to detect
the first echo. Therefore other echoes may affect when the
sensor is triggered again giving a false reading. Fig.2.21
clearly illustrates this scenario. So to avoid these false
readings sufficient amount of time should be left before
triggering the sensor again.
The minimum time the wave travels with echo power higher
than the threshold value was found to be 35ms. When we
have set the triggering time as greater than this value, the
output was stable and is shown in Fig.2.22.
Fig.2.22. Triggering time 35 ms Object at the point-
{x=0, y=0, z=300}
We have developed an algorithm in order to overcome the
problem of multiple echoes and false readings which is
described in section 2.5.
2.5 Generalized Filtering Algorithm
After a thorough analysis of this sensor, we were able to
develop a general filtering algorithm to eliminate the
uncertainty of the sensor data. The functionality of the
filtering algorithm can be summarized as follows.
Variables defined (assuming data is coming from m number
of sensors at n number of scans):
Input data at kth
scan� ri (k); i= 1,2,3,…m
k=t=f(∆T),∆T=1/ft
Output data array � Rm
Intimidate data array � Fm
Pointer array for one set of data � fm
2D array to indicate whole set of data � Pm, n
Pre defined range value � d1
A. If ri (k) = 0
Pi (k) = 0
Ri (k) =
B. Else
a. If Ri (k-1)- d1 < ri (k) < Ri (k-1) + d1
Pi (k) = 1
Ri (k) =( Fi (k-1) + ri (k) ) / 2
b. Else
Pi (k) = 0
I. If fi(k-1) = 1
fi(k) = 0
If Fi (k-1)- d1 < ri (k) < Fi (k-1) + d1
Ri (k) =( Fi (k-1) + ri (k) ) / 2
II. Else
fi(k) = 1
Fi(k) = ri(k)
Ri(k) = Ri(k-1)
Same process is executed for the scans, k+1, k+2, …
So the output of the algorithm is smoothened one set of data
which is reasonably able to represent all the n scans. Since it
False echoes
has correlated several set of data the filter has successfully
tracked the object for a period of time.
This filtering algorithm is shown in Fig. 2.23.
Fig.2.23. Flow Chart of the Filtering Algorithm
Behavior of raw data and filtered data is demonstrated in our
GUI using two graphs. Two axes of the graph indicate the
scanning time & distance, and at the same time 9 colors are
used to represent nine sensors. Fig.2.24. shows the behavior
of one sensor data, raw data and filtered data, at an instance
where the module is detecting an object at the distances of 2
m.
Fig.2.24. Graphs of the Raw Data and Filtered Data for one
sensor
3. CONCLUSION
This work provides the necessary guidelines for the
optimum usage of the most frequently used sonar range
finder SRF05.
Sonar beam pattern in 3D has been determined accurately
which would limit the failures due to the assumptions about
the beam pattern.
Best triggering frequency has been found to be 33ms. This
allows to trigger the sensor at a faster rate as required by the
application.
The ideas presented under sonar wave propagation and
multiple echoes provide a better insight to the behavior of
these sensors and would definitely be advantageous in
usage. The maximum angle of detection also provides the
necessary information in using these sensors to detect
inclined objects and to avoid missing the important data.
If these guidelines are adhered, there would be a much
better possibility to obtain the maximum usage of range
finder SRF05 sonar.
4. REFFERENCE
[1] Joanne walker, “Intelligent Robotics”
[2] SRF05 Technical Specifications
[3] Daniel .J. Dailey, Patricia Harn, and Po-Jung Lin “ITS
Data fusion,” Washington State Transportation
Commission, Washington, April 1996.
Uncertain
behavior
of the data
Smoothed
data