Automated Test System With Vision
Automated Test System With Vision
Automated Test System With Vision
Tan Wei Ren, Wan Shahmisufi bin Wan Jamaludin, Kueh Ying Lin, Muhammad Na-
siruddin Mahyuddin, and Bakhtiar Affendi Bin Rosdi
School of Electrical & Electronic Engineering Campus, Universiti Sains Malaysia, 14300 Ni-
bong Tebal, Pulau Pinang
lanson1991@hotmail.com, wanshahmisufi90@gmail.com,
kuehyinglin93@gmail.com, nasiruddin@usm.my, eebakhtiar@usm.my
Keywords: Needle gauge test, white pixel intensity test, automated instrument
cluster test (AIC).
1 Introduction
The instrument cluster is generally located directly above the steering wheel in every
vehicle. It is used to display important vehicle operation information to the driver such
as vehicle speed, fuel level, temperature of engine and status of various vehicular sys-
tems. A clear and accurate presentation of the vehicle condition is very important at any
given point of time to ensure the safety and comfort of the driver. The information are
usually in the form of needle gauges, digital readouts and lighted symbols which are
centrally in the primary viewing range of the driver.
2
Nowadays, most of the vehicle instrument cluster consist mainly of needle gauges
driven by stepper motors together with all of the warning lights that are built into a
single module [1]. In more luxurious car, more advance instrument clusters are used
which have some sort of digital multi-display which can be changed by inputs from the
driver to show different information about the vehicle and allow the driver to change
various settings.
The inclusion of machine vision in automotive manufacturing processes in the pre-
vious years as stated in [2] and [3] has been a revolutionary process that are able to
reduce inspection time as well as labor cost during the process of testing and validation.
The importance of including machine vision system in manufacturing process is that
the development cycle of a product can be reduced and the conventional manual testing
process which is less efficient will be replaced with highly efficient machine vision
system.
In [4-7], automation for the instrument cluster testing and validation using machine
vision had been developed using industry camera which has image processing software
that used library vision tools.
From existing testing, industry is using semi auto testing in the instrument cluster
which tester is needed to control the signal sending to cluster and spectating the result
by using human eyes. This may need a longer time for complete testing a cluster and
human error may occur.
In this paper, we proposed needle gauge test and signal indicator test to visual output
of instrument cluster. In our developed algorithm, we calculate the angle of the needle
to compute the result of needle gauge test and check whether the signal indicator light
is on or off in signal indicator test. With the proposed algorithm for needle gauge test,
pointer detection technique will get the angle of the needle and come out the result
based on the angle. Besides that, pixel intensity test will check the white pixel intensity
percentage in particular signal indicator light area. If the percentage of the white pixel
is exceed constant value then the algorithm will determine that the signal indicator light
is “ON”.
This paper consists of four main sections. Second section will explain about the al-
gorithm of computer vision which divided in two categories which are needle gauge
test and signal indicator test. Third section will show the result of this research. Lastly,
forth section will finalize this research.
In this paper, computer vision is used to capture and process the image of the Auto-
mated Instrument Cluster. Needle gauge test and signal indicator test are two main tests
in AIC test algorithm.
the system will compute the vehicle’s speed in kilometer per hour, km/h which show
on speedometer, the engine revolution per minutes (RPM) in tachometer, fuel level in
fuel gauge and engine temperature level in engine temperature gauge by getting the
angle of the needles.
The flow is start by capturing and calibrating the cluster’s image to make sure the
image is align correctly. Secondly, it will crop out the section of gauge which is going
to test. In [8] image thinning is used for analog measurement, while in this system im-
age thinning is used to reduce the noises on the blob and error when detecting the line.
Then, the cropped image is converted in to binary image to prevent noise for image
processing. After that, Hough transform line detection algorithm is applied to determine
the position of the pointer. Lastly, it will calculate the angle of the pointer and compute
the result of the particular gauge test (KM/H, RPM, Fuel level, Temperature level).
Needle gauge test will calculate based on formula given in Eq. (1):
Result = Constant * Angle. (1)
Constant in needle gauge test is determined by formula given in Eq. (2):
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡(𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑠𝑝𝑒𝑒𝑑−𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑠𝑝𝑒𝑒𝑑(20))
𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = (2)
𝐴𝑛𝑔𝑙𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2 𝑝𝑜𝑖𝑛𝑡𝑒𝑟 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛(𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑎𝑛𝑔𝑙𝑒−𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑎𝑛𝑔𝑙𝑒)
The algorithm to measure all the gauge are the same, the only difference is the area to
be cropped.
Fig. 1 shows the process of needle gauge test of speedometer at 80km/h, (a) is showing
the image that cropped out from whole cluster’s image. Then (b) is showing the cropped
image is been converted into binary image. After that, image thinning is applied on the
binary image. Follow by Hough transform is used to detect the needle in the image.
Lastly, system will calculate the angle between the reference line (20/km/h) and the red
line, based on the angle difference between two lines, system will compute the result
of current speed as in (e)
3 Result
Needle gauge test is test that used to measure the needle’s angle for speedometer, ta-
chometer, fuel gauge and temperature gauge. Based on the angle of the needle, those
result of gauge will be calculated and shown. In needle gauge test, we took 50 photos
of each gauge to test the algorithm. Those 50 photos is to test the deviation error occurs
in this algorithm. After test the algorithm and the algorithm is functioning well, we can
proceed to real time testing of the instrument cluster by using the algorithm.
In fig. 3, speed log file is a log file that show the result of speed after calculation from
the speedometer. In fig. 3, showing the image that captured in live (left side is first test
and second image is last test in the log file) and stored in particular folder and captured
speedometer image. Besides that, speeds is the speed that calculated based on the im-
age. Then the speeds in Txt is the expected speed that given by the test instruction in
text file format. Furthermore, status is to check that the speeds and speeds in Txt is
matched with stated tolerances.
Log file is generated once the testing is ongoing, all the result will be stored for re-
view after the testing. All needle gauge test no matter tachometer, fuel gauge, temper-
ature gauge will be same on the display but only background calculation is different.
4 Conclusion
In this research, the proposed algorithms are solving the inspection problem in auto-
mated instrument cluster test. Based on the result stated above, the proposed algorithms
can produce accurate results under a controlled environment with fix position of the
instrument cluster and surrounding lighting. Angle calculation is chosen and the pixel
intensity test is favored because it can get an accurate result of 2.3% and the processing
time requires less than 0.5s which is highly efficient.
In conclusion, the proposed algorithms for the automated testing are able to produce
desired result and can be implemented and replace the conventional manual testing in
the manufacturing line in the future.
References
1. Automotive Electronics - Instrument Clusters. Available: http://www.cvel.clem-
son.edu/auto/systems/instrument_cluster.html. 2016/05/12
2. P. Hage and B. Jones, "Machine vision-based quality control systems for the automotive
industry," Assembly Automation, vol. 15, pp. 32-34, 1995
3. S. Quan, X. Ning, and C. Yifan, "Development of an automatic optical measurement system
for automotive part surface inspection," in Proceedings, 2005 IEEE/ASME International
Conference on Advanced Intelligent Mechatrocnis., 2005, pp. 1557-1562.
4. S. Das and P. Das, "Competitiveness and its impact on research and development in indian
automobile industry," International Journal of Management and Transformation, vol. 5, pp.
79-89, 2011.
5. P. Dr.B.Sharmila and P. S. P.Venkadesan, "Automation for instrument cluster panel testing
using machine vision and NI-PXI hardware-in-the-loop," International Journal of Advanced
Research in Electrical, Electronics and Instrumentation Engineering, vol. 4, 2015.
6. G. Tsampardoukas and A. Mouzakitis, "Hardware-in-the-loop visual display validation uti-
lising vision system," in Control 2010, UKACC International Conference on, 2010, pp. 1-5.
7. H. Yingping, A. Mouzakitis, R. McMurran, G. Dhadyalla, and R. P. Jones, "Design valida-
tion testing of vehicle instrument cluster using machine vision and hardware-in-the-loop,"
in Vehicular Electronics and Safety, 2008. ICVES 2008. IEEE International Conference on,
2008, pp. 265-270.
8. J. Zhang, Y. Wang, and F. Lin, "Automatic reading recognition system for analog measuring
instruments base on digital image processing," Journal of Applied Sciences, vol. 13, p. 2562,
2013.