IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS

Research on Intelligent LED Headlamp System Based on Multi-Sensor

Fusion
To cite this article: Youyu Wu and Longqin He 2019 IOP Conf. Ser.: Mater. Sci. Eng. 677 032055

View the article online for updates and enhancements.

This content was downloaded from IP address 158.115.243.230 on 10/12/2019 at 17:16

IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Research on Intelligent LED Headlamp System Based on

Multi-Sensor Fusion

Youyu Wu*, Longqin Hea

Wuhan University of Technology, Wuhan, China

*Corresponding author e-mail: wuyouyu1@whut.edu.cn, ahelongqin1947@qq.com

Abstract. This paper researches on automobile headlamp system. Multi-sensor fusion

technology is used in the system, which can realize for data acquisition of full
automation, including the road conditions and the vehicle’s own operating status. On
this basis, the deflection angles of headlamp at bend and ramp are calculated in real time,
and the illumination angle and the illumination distance of the headlamp are adjusted
accordingly. Finally, the research on intelligent LED headlamp system with the
functions such as intelligent switching of low beam and high beam, bend adaptive
lighting and ramp adaptive lighting is completed. The test results show that the system
performance meets the basic requirements of automobile adaptive lighting.

1. Introduction
With the improvement of people's consumption ability and the rapid development of science and
technology, more and more people pay attention to the driving safety and environmental protection of
vehicles. As one of the important parts of future automobile driving plan, headlamp is also invested by
more and more automobile companies and research institutions.
In the world, the intelligent research on automobile headlamps began in 2003. In 2005, HELLA
announced that it has produced the AFS system, in which the headlamps can be automatically adjusted
based on the road conditions and the external environment [1]. In 2013, new Audi A8L was equipped
with matrix LED headlamps with dynamic steering function realized by changing the lens and reflector
to illuminate the bend ahead of time [2, 3]. In 2015, laser headlamp technology, which not only realizes
the functions of the traditional AFS, but also makes the headlamp illumination distance from 300 meters
to 600 meters, was demonstrated on a new generation of BMW 7 Series vehicles codenamed G11/G12
[4]. There is also an AFS system based on GPS positioning system developed by Visteon, the headlamps
are adjusted in time through precise GPS positioning and map data [5].
China's intelligent headlamp research started later than the international. In early 2007, Shenyang
North Automobile Headlamp Auto-steering Factory successfully developed automobile automatic
steering headlamps, which broke the monopoly of foreign companies in this respect [6]. In July 2009,
the new Lacrosse equipped with a follow-up xenon headlamp developed by HiRain Technology, was
officially launched. The headlamp has the functions of following-up turning at bends, automatically
enhancing near light intensity at low speed and automatically increasing light distance at high speed [7].
In the same year, Jilin Dongguang Ruibao Automobile Lamp Co., Ltd. successfully developed the first
full LED headlamp in China [8].

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

The above are the research status at home and abroad. In summary, at present, the research can be
regarded as three operations for high beam and low beam, including on-off control, brightness
adjustment and direction adjustment.

2. System structure

2.1. System function description

The intelligent research on automobile headlamp in this paper is divided into three parts as follows.
⑴ Intelligent switching of low beam and high beam
After the light AUTO key is pressed, the headlamp can be automatically turned on or off according
to the ambient illumination. With the headlamp turned on, the high beam and low beam can be
automatically switched according to the speed of the vehicle.
⑵ Bend adaptive lighting
By real-time acquisition of steering wheel angle and vehicle speed, the illumination angle of the
headlamps is automatically adjusted, and the light intensity inside the curves is enhanced.
⑶ Ramp adaptive lighting
The illumination distance of the headlamps is automatically adjusted based on real-time information
such as vehicle speed and body inclination.
There are two requirements for the opening of bend adaptive lighting and ramp adaptive lighting.
One is that the AUTO key of the headlamp has been pressed, and the other is that the headlamp has been
turned on.

2.2. System structure

According to the three functional requirements of automobile intelligent headlamp, there are some
requirement analysis of hardware system as follows.
The intelligent LED headlamp system based on multi-sensor fusion consists of three key points: 1)
LED headlamp; 2) AFS system; 3) multi-sensor fusion.
The first and second key requires that the hardware platform needs an LED headlamp with low beam,
high beam and a mechanical rotation structure that adjust the beam in two axes. The hardware platform
is also required to have LED drive unit and motor drive unit respectively. As for the third key, there are
many sensors needed by the hardware platform, including photosensitive sensor, speed sensor, steering
wheel angle sensor and six-axis sensors, and possibly image sensors. The control and data
communication between the driving units and the sensors require that there is main control unit, referred
to as MCU, needed in hardware platform.
Photosensitive Steering Wheel Vehicle Speed
MPU-6050
Sensor Angle Sensor Sensor

Control Unit and

Driving Units

Headlamp with two-

axis rotating structure

Figure 1. The hardware platform framework.

2
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Figure 1 shows the hardware platform framework of this paper. Figure 2 shows the communication
relationship between MCU, driver units, sensors, motor and LED headlamp.

DC +5V M C U , Six-Axis Sensor in Assembly Circuit ， C a r Body S ix-axis

Sensor, Photosensitive Sensor
DC +8V Horizontal Motor Drive Unit, Vertical Motor Drive Unit, Steering
Power Wheel Angle Sensor, Speed Sensor
DC +9.4~+26V
LED Headlamp Driver Unit

LIN Horizontal Motor DRIVE

Horizontal Motor
Drive Unit

LIN Vertical Motor Drive DRIVE

Vertical Motor
Unit
Assembly Circuit Board

CAN
Steering Wheel Angle Sensor
PWM
Speed Sensor
MCU I2C
Photosensitive Sensor
I2C
Car Body Six-axis Sensor

I2C Six-Axis Sensor in

Assembly Circuit
SPI DRIVE
LED Headlamp Near-light LED Light Set
I/O Driver Unit DRIVE
Far-light LED Light Set

Figure 2. Working block diagram of hardware system.

The analog input of hardware system includes body inclination. The digital inputs include ambient
illumination, vehicle speed, steering wheel angle, etc. The digital outputs include horizontal motor
adjustment state, vertical motor adjustment state, near-light intensity, far-light intensity, near-light
extinction, far-light extinction. There is no analog output in this hardware system. There is a LIN
communication interface, a CAN communication interface, a serial peripheral interface, three analog
I2C communication interfaces and a pulse capture interface in the MCU with MC9S12P128 as the core.
From the requirement analysis of the hardware system, to the selection of components, to the design
of the assembly circuit board, then to the installation and debugging of the hardware platform, the
hardware platform of the intelligent LED headlamp system based on multi-sensor fusion is completed
finally.

3. System control strategy

3.1. System working constraints

The working constraints of intelligent LED headlamp system based on multi-sensor fusion are as follows:
⑴ The horizontal adjustment range of automobile headlamp is that the left headlamp adjustment
angle is -15°~8°, and the right headlamp adjustment angle is -8°~15° [9, 10].
⑵ Conditions for headlamps to be turned on are that ambient illumination is less than or equal to
200 lux, and detailed data analysis is shown in Figure 9.
⑶ Relevant traffic regulations for the intelligent switching of high beam and low beam in the case
of low visibility [11]:
1) Within 150 meters from the vehicles coming in the opposite direction, the vehicle is required to
change to low beam instead.
2) When meets with non-motorized vehicles on narrow roads or narrow bridges, the vehicle is
advised to use low beam.

3
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

3) When the vehicle is driving at close range in the same direction with other vehicles, no high beam
shall be used.
4) When passing through sharp bends, slopes or intersections without traffic lights, the vehicle is
recommended to alternately use high beam and low beam to indicate.
⑷ In the case of low visibility, it is recommended to use a low beam when the vehicle speed is lower
than 30 km/h, otherwise the high beam can be used. Vehicles are recommended to use low beam at night
through a section with good lighting conditions.

3.2. System Control Strategy

The three important research stages of this system are: 1) building the hardware platform environment;
2) formulating the system control strategy and compiling the bottom program and control program of
the software system; 3) verifying the feasibility of this system in accordance with the national standard
of automobile headlamp and the working constraints of this system. This section mainly introduces the
control strategy of the system.

Start

Key shift to ON

System Fault Detection and Alarm

Processing Module

Has the AUTO key N

been pressed?

Y Ambient illumination
greater than 200 Lux

N
The intelligent switching module
of far and near light

Bend adaptive lighting module

Ramp adaptive lighting module

N
Key shift from ON to OFF

Y
Horizontal motor and vertical motor
restored to initial position

Exit system

Figure 3. Working block diagram of hardware system.

The main program flow chart of this system is shown in Figure 3. The main program workflow
mainly includes system fault detection and alarm processing module, the intelligent switching module
of high beam and low beam, bend adaptive lighting module and ramp adaptive lighting module.
The working principle of the system will be described in the following two paragraphs.

4
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

There are three initialization steps before starting the system. Firstly, checking whether the car key
is in ON gear is needed. Secondly, the headlamp fault detection is needed, and if the faults are detected,
the alarm will be issued and the system will exit. Thirdly, it is necessary to detect whether the driver has
pressed the AUTO key of the lamp. If not, it means that the system is not started.
After the system is started, first of all, the ambient illumination is automatically detected. If the
illumination is not greater than 200 lux, the headlamps are automatically turned on. Next, the intelligent
switching module of high beam and low beam is entered, and high beam and low beam are switched
according to the speed of the vehicle. Then, the bend adaptive lighting module is entered, the horizontal
motor drive the lamp to the position which is calculated according to the specific bend follow-up
algorithm, and the light intensity inside the curves is enhanced in the same time. After that, the ramp
adaptive lighting module is entered, the vertical adjustment angle of the lamp is calculated with the
specific ramp adjustment algorithm, and the illumination distance of the headlamps are adjusted
accordingly. Finally, the above process is cycled until the car key is in the OFF position. If the driver
turns the car key to the OFF position, the system will exit after the motor returns to the home position.

Start

Vehicle speed Y Switch to the

greater than 55km/h far lights

Vehicle speed N Switch to the

greater than 30km/h near lights

Switch to the far lights

Allow manual adjustment to the near

light according to road conditions

End

Figure 4. Flow chart for the intelligent switching module of far and near light.

Start

Does the steering wheel N

have a deflection angle?

Y
Calculating lamp deflection angle
based on bend steering algorithm

Horizontal motor adjusts beam

direction in real time

Enhance the light intensity

inside the bend

End

Figure 5. Flow chart for bend adaptive lighting module.

5
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Start

Slope inclination N
greater than 5°

Y
Calculating vertical adjustment angle
based on slope adjustment algorithms

Vertical motor adjusts headlight

illumination distance in real time

End

Figure 6. Flow chart for ramp adaptive lighting module.

The details of the system fault detection and alarm processing module is shown in Table 4. The flow
chart for the intelligent switching module of high beam and low beam is shown in Figure 4. The flow
chart for the bend adaptive lighting module is shown in Figure 5. And the flow chart for the ramp
adaptive lighting module is shown in Figure 6.

4. Experimental result
In this system, the NCV78663 chip is used as the main chip of the LED headlamp driving unit. The two
schemes of adjusting headlamp brightness are: 1) setting dimming duty factor of LED strings; 2)
controlling the current setting ratio of LED strings. In this paper, we test the data for these two schemes,
and the obtained data is fitted by first-order linear fitting. The test object is 1W lamp beads.
The current and voltage parameters of each lamp bead are measured by changing the number of lamp
beads in series or the dimming duty factor while the other parameters remain the default state. The
dimming duty factor can be abbreviated as DDF. The measurement data are shown in Table 1, and the
relationship between the working parameters of the beads and the DDF is shown in Figure 7.

Table 1. Measurement data by changing the number of lamp beads or the DDF.
Lamp beads*1 Lamp beads*2 Lamp beads*3 Lamp beads*4
DDF
V/V I/mA V/V I/mA V/V I/mA V/V I/mA
0.00% 2.61 20.09 5.39 43.50 8.12 40.79 10.84 42.62
10.00% 2.59 18.16 5.33 39.17 8.05 39.39 10.75 38.24
20.00% 2.58 16.60 5.30 35.61 8.00 33.29 10.68 34.91
30.00% 2.56 14.75 5.24 30.76 7.92 28.87 10.58 30.12
40.00% 2.54 12.10 5.20 25.89 7.85 24.30 10.48 25.45
50.00% 2.52 11.90 5.14 24.40 7.76 23.90 10.39 23.60
60.00% 2.50 8.04 5.08 16.16 7.69 15.21 10.28 15.80
70.00% 2.47 5.48 5.03 11.52 7.61 10.72 10.18 11.40
80.00% 2.45 3.48 4.97 6.81 7.53 6.32 10.07 6.40
90.00% 2.44 2.03 4.92 3.08 7.46 2.97 9.99 3.11
94.14% 2.43 1.34 4.91 1.71 7.44 1.43 9.96 1.53
94.53% The lamp beads can't be turned on.

6
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Figure 7. Output and the DDF.

As shown in Figure 7, when the number of 1W beads in series is greater than 1, the relationship
curves between the output parameters and DDF basically coincide. With the increase of DDF, the beads’
current decreases linearly and rapidly, and the voltage decreases in a very small range.
Changing the number of lamp beads in series or the current setting while the other parameters remain
the default state is to measure the data of the current and voltage parameters. The obtained data are
shown in Table 2, and the relationship between the working parameters of the beads and the current
setting is shown in Figure 8.
The relationship curves between the output parameters and the current setting basically coincide.
With the increase of current setting ratio, the output current of the beads increases linearly and rapidly,
and the voltage increases in a small range.
Table 2. Measurement data by changing the number of lamp beads or the current setting.
Lamp beads*1 Lamp beads*2 Lamp beads*3 Lamp beads*4
Current setting
V/V I/mA V/V I/mA V/V I/mA V/V I/mA
0.00% 2.64 22.4 5.46 49.3 8.10 48.2 10.80 48.1
20.00% 2.89 176.0 5.88 259.0 8.67 262.4 11.70 257.7
40.00% 3.10 432.0 6.17 459.0 9.24 459.0 12.42 459.0
60.00% 3.28 682.0 6.48 678.0 9.62 670.0 12.82 664.0
80.00% 3.41 898.0 6.61 837.0 9.67 803.0 -- --
100.00% The lamp beads burned quickly.

Figure 8. Output and the current setting.

7
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Table 3. Measurement data of ambient illumination at different times.

lux ratio lux ratio lux ratio lux ratio
1 287 0.50 617 0.25 747 0.30 247 0.50
2 169.2 0.50 452 0.40 385 0.45 207 0.65
3 156 0.75 360 0.30 312 0.35 157.2 0.55
4 114.6 0.80 320 0.30 265 0.40 133.2 0.55
5 77.3 0.85 168.1 0.35 159.3 0.55 101.2 0.65
6 76.6 0.75 132.5 0.50 121.9 0.50 77.6 0.65
7 62.1 0.90 116.3 0.65 89.7 0.55 59.6 0.85
8 53.9 0.75 102.6 0.50 66.9 0.70 33.8 0.95

In the research of intelligent automobile headlamp system, the threshold of ambient illumination is
an important factor affecting system quality. Therefore, we measured and counted the headlamp opening
situation and ambient illumination at dusk for 5 days. For each calculation of the headlamp opening
situation of 20 vehicles, we records the ambient illumination at that time. After screening out the data
with larger errors, the remaining reasonable data are shown in Table 3.
Fitting the data with a third-order polynomial, the relationship curve is obtained as shown in Figure
9. The expression of the relationship between the ratio of car lights on and the ambient illumination can
be listed as equation ⑴.
y  2976.92  x 3  6773.55  x 2  5321.44  x  1531.20 ⑴

Generally, the proportion of automobile lights on that reaches more than 1/2, means that the ambient
illumination at that time is the threshold of the system headlamp on. By substituting x = 0.5 into equation
⑴, the result is y = 191.76 lux. Therefore, this paper takes 200 lux as the threshold of ambient
illumination.

Figure 9. The Relationship between illuminance and headlamp turn-on ratio at different times.

In this paper, a closed-loop feedback control system is composed of LED driver, motor drivers,
sensors and MCU. There are two fault detection mechanisms in the system, one is active feedback
mechanism, and the other is passive feedback mechanism.
The active feedback mechanism of fault detection refers to actively using a specific I/O to feed back
its fault information to MCU after a fault occurs in the drivers or sensors. The passive feedback

8
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

mechanism of fault detection refers to that when it’s necessary to know whether a driver or sensor has
a fault, the MCU sends corresponding instructions to the fault detection object by in a specific
communication protocol. After receiving the instructions, the detecting object feeds back the fault
information to the MCU. In this system, the active feedback mechanism is used as the fault detection
mechanism of hardware units such as LED driver, steering wheel angle sensor, speed sensor, etc. And
the passive feedback mechanism is used as the fault detection mechanism of the hardware units
including LED driver, motor driver and photosensitive sensor.
The fault detection types and treatment measures of the system are briefly summarizes in Table 4.
As we see, there are three levels of alarm mechanism in Table 4 for fault diagnosis. The treatment
measure of level 1 alarm is that red warning light is turned on and the system is forced to exit, as well
as the car is recommended to park overhaul immediately. The treatment measure of level 2 alarm is that
orange warning light is turned on, the system is forced to exit and the car is recommended to repair when
idle. The treatment measure of level 3 alarm is that the yellow warning light is turned on, but the system
is still working, and the car is suggested to repair when idle.

Table 4. Fault diagnosis types and treatment measures.

Alarm Level Fault Diagnosis Types Treatment Measures
1 Level 1 alarm LED string open circuit Parking overhaul
2 Level 1 alarm LED string short circuit Parking overhaul
3 Level 1 alarm Individual LED shorted Parking overhaul
4 Level 1 alarm LED string over current Parking overhaul
The LED driver communicates abnormally with the
5 Level 2 alarm Exit system forcibly
MCU
6 Level 2 alarm Horizontal motor or vertical motor works abnormally Exit system forcibly
7 Level 2 alarm Photosensitive sensor works abnormally Exit system forcibly
8 Level 2 alarm The steering wheel angle sensor works abnormally Exit system forcibly
9 Level 2 alarm Speed sensor is working abnormally Exit system forcibly
10 Level 2 alarm Six-axis sensor works abnormally Exit system forcibly
11 Level 3 alarm LED driver booster is under voltage Indicator light alarm
12 Level 3 alarm LED driver issues a thermal warning Indicator light alarm
13 Level 3 alarm LED driver booster reaches maximum duty cycle Indicator light alarm

5. Conclusion
In this paper, the multi-sensor fusion technology is adopted in this intelligent LED headlamp system.
Acquiring data such as the road conditions and the vehicle’s own operating status at the same time by
multiple sensors, improves the sensitivity of the system. Compared with most intelligent headlamp in
the current market, this system adds the ramp adaptive lighting function, which enables the vehicle to
automatically adjust the illumination distance of the headlamp at the ramp. The ramp adaptive lighting
function improves the intelligence and security of the system. All in all, this paper completes the research
and implementation of intelligent LED headlamp system based on multi-sensor fusion, which increases
the reliability and efficiency of the system.

Acknowledgments
Thanks to Pengpeng Yao and Yulong Wang for their contributions in this project, and thanks for the
support from teacher Xiaofang Du.

References
[1] Zhang Xiaoming. “The Design and implementation of Adaptive Front Lighting System” Harbin
Engineering University, 2014.
[2] LUO Dezhi, NIU Pingjuan, et al. “The Present Situation and Development of Intelligent

9
IMMAEE 2019 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 677 (2019) 032055 doi:10.1088/1757-899X/677/3/032055

Headlight of Automobile” China Illuminating Engineering Journal, 2017, vol.28, no.5, pp.72-
78.
[3] TAN Xitang, LIU Sha, et al. “Matrix LED high-beam intelligent assistant control system” Journal
of Computer Applications, 2019, vol.39, no.6, pp.1855-1862.
[4] Liu Peng. “Heat Dissipation Analysis and Optimization of LED Heaplamp” Wuhan Institute of
Technology, 2014.
[5] Wang Jinlong. “Analyse and Research of Adaptive Front Light System in Automobile” Harbin
Institute of Technology, 2014.
[6] Renzhong Huang. “Development of Adaptive Lighting System in Automobile” Dalian University
of Technology, 2008.
[7] Xiaojing Gao. “Study on Front-Lighting Systems (AFS) of Motor Vehicles” Beijing Institute of
Technology, 2016.
[8] Wei Hui. “Research on Light Distribution Technologies of Vehicle LED Adaptive Front-lighting
System Headlamp Based on the Energy Utilization” South China University of Technology,
2012.
[9] WANG Weijie, LIU Qiulin. “Research on the Mathematical Model of Bending Mode of Adaptive
Front-lighting System” Machinery & Electronics, 2015, no.10, pp.7-11.
[10] Liu Qiulin. “Research on the Adaptive Front Lighting System Based on Driving Environments”
Harbin Institute of Technology, 2015.
[11] China’s State Council. “Regulation on the Implementation of the Road Traffic Safety Law of the
People’s Repuplic of China” China, 2017.

10