939218

Research ArticleA Study of Bending Mode Algorithm of Adaptive Front-LightingSystem Based on Driver Preview Behavior

Zhenhai Gao1,2 and Yang Li1

1 State Key Laboratory of Automobile Simulation and Control, Jilin University, Changchun 130022, China2 State Key Laboratory of Vehicle NVH and Safety Technology, Changan Automobile Holding, Chongqing 401120, China

Correspondence should be addressed to Yang Li; [email protected]

Received 25 February 2014; Accepted 4 April 2014; Published 4 May 2014

Academic Editor: Hamid Reza Karimi

Copyright 2014 Z. Gao and Y. Li. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The function of adaptive front-lighting system is to improve the lighting condition of the road ahead and driving safety at night.Thecurrent system seldom considers characteristics of the drivers preview behavior and eye movement. To solve this problem, an AFSalgorithmmodeling a drivers preview behavior was proposed. According to the vehicles state, the drivers manipulating input, andthe vehicles future state change which resulted from the drivers input, a dynamic predictive algorithm of the vehicles future trackwas established based on an optimal preview acceleration model. Then, an experiment on the change rule of the drivers previewdistance with different speeds and different road curvatures was implemented with the eye tracker and the calibration method ofthe drivers preview time was established. On the basis of these above theories and experiments, the preview time was introduced tohelp predict the vehicles future track and an AFS algorithm modeling the drivers preview behavior was built. Finally, a simulationanalysis of the AFS algorithmwas carried out. By analyzing the change process of the headlamps lighting region while bend turningwhich was controlled by the algorithm, its control effect was verified to be precise.

1. Introduction

AFS (adaptive front-lighting system) is a front-lighting sys-tem that can change the light pattern and illumination areaaccording to the vehicles state such as the velocity, thesteering wheel angle, and road environment to light the roadahead effectively so as to reduce accidents at night. Accordingto AFS productions of international companies and researchinstitutes research findings and directions currently, theclassic algorithm of AFS can be divided into the followingfour categories.

(1) The AFS Algorithm Based on the ECER123 Regulation[1]. The basic demand of this algorithm was to control thedeflection angle of the headlamp at an allowed range of theregulation. Strambersky et al. (European Lighting Technol-ogy Centre of Visteon Company) used Ackerman steeringgeometry principle to calculate the headlamps deflectionangle in the regulations allowed range [2]. The algorithm

based on the regulation needed to be improved in accuracyand trajectory prediction.

(2) The AFS Algorithm Based on Brake Safety. The algorithmused the stopping sight distance to calculate arc length andthe stopping sight distance was larger than brake distance tomake sure that the driver would observe the obstacle in front.Huang, Lifu and Qian, Yu, Deng, Mclaughlin et al., and Ronget al. had used similar methods in their studies [38]. Thedesign based on brake safety could ensure the safety, but theheadlamps lighting location had a lag relative to the vehicleslocation.

(3) The AFS Algorithm Based on Early Lighting. When a carwas driven on the curve, the headlamp should be lighted onthe location where the car would arrive at after a while. Thiskind of research solved the problem to some extent that theheadlamp illumination direction had a lag compared to thedriving direction [9, 10].This kind of AFS can be divided intononpredictive AFS and predictive AFS.

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014, Article ID 939218, 11 pageshttp://dx.doi.org/10.1155/2014/939218

2 Mathematical Problems in Engineering

(a) Nonpredictive AFS. Koji Ishiguro and Yuji Yamada whowere in Denso and Toyota in Japan studied the relationshipbetween vehicles speed and drivers fixation distance duringbend driving in the day time.They concluded that the fixationdistance got larger while the speed increased. By studyingdrivers reaction time in dangerous circumstances such ascollision, they regarded = 3 s as an appropriate time whichthey used in their AFS algorithm [11]. Nonpredictive AFScould improve the lighting condition of the road ahead tosome extent, but it could only use the current steering wheelangle and velocity to estimate the vehicles transient currenttrajectory and then calculate the headlamps deflection angle.When the roads curvature changed a lot, the beam mightdeflect too slow in the corner and even deflected to a wrongdirection at an S-curve.

(b) Predictive AFS. By GPS, electronic map, CCD which per-ceived the traffic environment information including vehiclelocations, road types, lane numbers and road curvatures,and vehicles state information which was used to predictcars future track, the P-AFS controlled the headlamp inadvance. Ibrahim of Visteon applied for the patent whichwas called Predictive adaptive front lighting algorithm forbranching road geometry, Patent number US 7558822 B2[10]. They matched the GPS information with the databaseof electronic map to confirm the vehicles current locationand predict the roads information such as road type, lanenumbers, and curvature and road branches and finally cal-culated the headlamp deflection angle by P-AFS algorithm.Kim et al. in Korea (2010) proposed a new method of P-AFS which was called curvature estimated swivel in theirpaper [12]. By using the existing devices such as LDWs toreplace the GPS to predict road curvature, the headlampsdeflection angle was calculated. Predictive AFS solved thenonpredictive AFSs error problem to some extent when theroads curvature changed a lot, but it did not take the driversvisual characteristics into consideration.

(4) The AFS Algorithm Based on the Drivers Vision in theCorner. Many researchers have studied the drivers sightmovement in a bend or crossing [1315]. Although theseresearches aimed to study the drivers fixation behavior whilenot providing information for the development or design ofAFS, they had some guiding significances for the design ofAFS.

On the basis of these experiments, Young et al. used theregular pattern of heads turning angle at daytime and nightto guide the design of AFS [16], but the paper did notmentionthe specific algorithm of AFS and the precision could not beensured because of the limit of measuring devices.

Considering the regulation and safety, the algorithm ofAFS currently estimated the vehicles future driving trackaccording to vehicles moving states, traffic environment,and drivers wheel input. Then the AFS algorithm controlledthe headlamps deflection angle to illuminate the estimateddriving track in advance. But most of these algorithmsassumed that the speed and thewheel angle would not changeand assumed the driving track as a circle. According to Zhanget al.s study on vehicles lateralmoving characteristics [1719],

the possible variation of vehicles longitude and lateral speedwas not considered which would also result in the change ofdriving track. In some circumstances such as in a curve entry,in a curve exit in an irregular road, or the drivers accelerationor deceleration to pass a varying curve, the algorithm couldnot predict the future driving track well and the effect of AFSwas not very well.

Meanwhile, as the most important active safety device,the headlamps main effect is to provide illumination for theroad ahead. Hence, the AFS should satisfy the safety demand,ensure that the drivers concerned region is fully illuminated,and must not influence the drivers fixation behavior. Itsillumination effect also has a direct relationship with thedrivers eye comfort. Compared to a good environmentalillumination condition such as in the daytime, it shouldnot additionally increase the drivers visual fatigue. In thisaspect, although some researches have used the statistic ruleof drivers visual field to control the headlamps deflection, inthe practical application, thismethodwhichwas totally basedon the statistic rule cannot ensure the safety.

Based on the considerations above, an AFS algorithmconsidering the drivers preview behavior was proposed.The vehicles kinematics and dynamics characteristics wereused comprehensively to calculate the vehicles future trackwhich was more reliable than the method purely basedon the vehicles state information the headlamps deflectionlag and direction error was avoided. Under the premise ofensuring safety, parameters of the drivers fixation behaviorswere introduced to increase the measuring precision of thedrivers visual statistical rule by using the eye tracker andthe headlamps illumination is more in compliance with thedrivers fixation behavior. This paper proposes an originaltechnology route; the work was first carried out in early 2010,and it has obtained national invention patents [20].

2. AFS Algorithm Considering the DriversPreview Behavior

2.1. AFS Algorithm. By modeling the drivers preview behav-ior at a bend, the vehicles future track in a period of timewas predicted according to the vehicles current state and thevehicles steady state response characteristics to the driverswheel, gas pedal, or brake pedal input. Simultaneously, thedrivers fixation location on the future track was determinedaccording to drivers preview behavior rule (the rule offixation location or preview spot) at real bends; then the head-lamps deflection angle was controlled and the location wasilluminated effectively. The technical route of the algorithmwas shown in Figure 1.

The algorithm simulated the drivers preview behavior ata large curvature bend and an integral algorithm to predictthe vehicles track was proposed. The algorithm was basedon the hypothesis of steady preview and dynamic correction[21]. According to the vehicles current state and its possiblechange which resulted from the drivers inputs, the vehiclesfuture track was predicted by the algorithm [22]. As Figure 2showed, the vehicles coordinate system at current time wasthe reference coordinate system. The preview time was

Mathematical Problems in Engineering 3

Preview distance field testing

Preview empirical formula

Steering wheel angle The deflection angle of the light typePredicting algorithm

of future trajectoryPreview

point

The bending mode algorithm of AFS withthe simulation of driver preview behaviorLongitudinal/lateral

vehicle

Throttle/brake pedalpressure

Figure 1: The specific technical route.

0

1

J

x0

y0

x1

y1

xjyj

Figure 2: The coordinate system.

evenly divided into pieces ( = /). During the previewtime, the longitudinal and lateral acceleration which resultedfrom the drivers wheel or gas pedal input were the same.

Because every time piece is quite short, the mutual effectof the vehicles longitudinal and lateral movement can beneglected.On the basis of the vehicles state (location, velocity,and acceleration) at the initial moment of the time piece,the vehicles state at the end of the time piece was calculatedusing the rigid body kinematics principle [23].Then from themoment , every time pieces vehicle state could be calculated.As (1) showed (at the time piece , the initial moment is 1and the endmoment is ), the combination of the location atevery time formed the vehicles future track in time:

[

] = [1

1

] + [

] ,

[

] = [1

1

] + 1,0 ([1

1

] +1

2[

] 2

) .

(1)

The coordinate transformation matrix is

0 = (cos sinsin cos

) . (2)

is path angle which is calculated by

= 1 +

. (3)

x

y

0

Predicting future trajectoryStarting place (x0, y0)

Preview point (xJ, yJ)

Figure 3: Adaptive front-lighting system deflection angle calcula-tion mathematical model.

= 1, 2, . . . , ; (, ), (, ), and (, ) were thevehicles location, the vehicles velocity, and the vehiclessteady acceleration at time ; was the vehicles course angleat time . The vehicles coordinate system at current time (0) was the reference coordinate system. During the previewtime, the longitudinal and lateral acceleration which resultedfrom the drivers wheel or gas pedal input were the same.

Then the vehicles mass centre coordinate (, ) after thepreview time can be calculated at the coordinate systemof current time . As Figure 3 showed, the point was thedrivers preview location at current time and the headlampsdeflection angle which was the angle between the vehiclespredicted location and the vehicles longitudinal axis wascalculated by

= arctan 0

0

. (4)

Based on the above theory, the measurement of the pre-view distance for real drivers was conducted. By the advancedeye tracker, the drivers fixation behavior was recorded whenthe driver drove through the bend. The drivers gazingdirection was analyzed and the average value of the previewdistance was calculated. Then an empirical equation of therelationship between preview distance, preview time andvelocity, and road curvature was proposed and the previewtime for the AFS algorithm was modified.


Table 1: The experiment velocity and the curvature radius.

Radius of curvature (m) Velocity (km/h)20 10, 15, 20, 25, 3030 20, 30, 4040 20, 30, 40

Figure 4: Drivers gaze point position comparison chart whenbeing close to the bend.

3. The Experiment Measurement and ChangeRule Study on the Drivers Preview Distance

3.1. Experiment Measurement

3.1.1. Experiment Equipment. SmartEye Pro eye tracker(sample frequency: 60Hz), A JETTA car made by FAW-Volkswagon, a 12V spare battery, meter ruler, the adhesivetape to make the lane, and others.

3.1.2. Experiment Conditions. Eight drivers with normalvision. Three quarter-circle lanes with curvature radius of20m, 30m, and 40m were used as the experiment road.With every lane, the experiment was done for three timesrepeatedly. Before the formal experiment, every driver did 10-minute exercise. The experiment velocity and the curvatureradius were shown in Table 1.

3.1.3. The Definition of Fixation Location. The SmartEye Proeye tracker recorded the drivers raw eye movement data.According to the thresholds of eye movement parametersset up by the users, the fixation and saccade behaviors wereconfirmed. Meanwhile, by the corresponding software, thevideo of the driving scene was replayed and the fixationlocation (the green circle in Figure 4) was spotted on thedriving scene.

The threshold settings of eye movement were as follows:

(1) the threshold of visual angle deviation: 2 deg.;(2) the eyeball movement velocity threshold during fixa-

tion: 15/s, which was the highest velocity of eyeballmovement during a fixation;

(3) the eyeball movement velocity threshold during sac-cade: 35/s, which was the lowest velocity of eyeballmovement during a saccade;

(4) the fixation duration threshold: 200ms.

According to the road environment, the state change ofthe vehicle, the fixation location (visual angle), and its change

(a)

(b)

Figure 5: Drivers gaze point position comparison chart before andafter into the corner: (a) is before into the corner, and (b) is after.

in the driving scene video, the entire driving process wasdivided into four sections: straight lane, being close to thebend, entering the bend, and being out of the bend.

Straight lane: in this section, the entire bend was shownin the drivers visual field. Repeated experiments showedthat the driver would sweep the bend and form a roughimpression of the available track; then the fixation locationwould be a few meters away from the vehicle as shown inFigure 4.

In the section of being close to the bend, because thevehicles movement state had not changed, the nonpredictiveAFS algorithm could not predict the bend ahead and activatethe AFS bend lighting function.

Entering the bend: as shown in Figure 5 (there werethree frames intervals between two images which meant thetime interval was 0.05 s). There was a saccade in almost allexperiments when the driver just drove from the being closeto the bend section to the entering the bend section. It wasalso shown in the drivers eye movement data that there was asaccade in the moment.This saccade behavior was defined asthe boundary between the being close to the bend sectionand the entering the bend section. After this saccade, thelane area was defined as the entering the bend section.

In the entering the bend section, the driver needed tosteer and the AFS needed to activate bend lighting functionand control the headlamps deflection according to somerules. So according to the purpose of this experiment and ouralgorithms parameters need, this sectionwas the emphasis ofour study.

Being out of the bend: the driver would brake when hewere going to be out of the bend. This section would not beanalyzed in detail.

3.1.4. The Preview Distance Calculation and the Analysis of ItsChanging Rule. The preview distance was calculated in the


Figure 6: The eye tracking systems installation location.

Gaze direction

O X

Y

Z

Horizontal plane

Vertical plane

Figure 7: The world coordinate system and gaze direction angle.

entering the entrance section during the driving process.The fixation point in the front near the centre line of theroad was defined as the preview point. The SmartEye Prosystem was installed as shown in Figure 6. The definition ofthe SmartEye Pro world coordinate system was as follows:the connection line between number 2 camera and number3 camera in SmartEye Pro system was the -axis of theworld coordinate, the midpoint of the connection line wasthe coordinate origin, and the-axis was in the vertical plane.The coordinate system was shown in Figure 7.

As shown above, the angle between the gaze directionsprojection in XOZ plane and the negative direction of theworld coordinates-axis (the positive direction of the vehiclecoordinates direction) was the gaze directions horizontaldirection angle. The angle between the gaze direction andthe horizontal plane was the gaze directions vertical angle.

The gaze directions horizontal angle is

= arctan GazeDirection.GazeDirection.

. (5)

The gaze directions vertical angle is

= arcsin GazeDirection.. (6)

GazeDirection.(, ) was the projection of the unitvector along the gaze direction on the (, ) axis of theworld coordinate system.

According to the gaze direction and the gaze origin inthe world coordinate system, the preview points coordinatecould be calculated in the world coordinate system as shownin Figure 8. The distance away from the driver which was thepreview distance could be calculated by (7) (the vehicles roll

Gaze position

X

Y

Z

Vertical plane

O

h

Previewpoint

D

Gaze origin

E1

E2

dpE

G

Figure 8: Preview distance calculation schematic.

Table 2: Preview distance data under 40m bend radius.

Data operation Bend radius (20m)10 km/h 15 km/h 20 km/h 25 km/h 30 km/h

Mean 9.6253 9.7248 9.8515 10.0431 10.1059Variance 1.0292 0.9282 0.8827 0.8319 1.2074Maximum 7.5492 7.7634 8.0599 8.0124 8.2775Minimum 12.3558 12.0545 12.1912 12.4505 12.1334

Table 3: Preview distance data under 20 km/h velocity.

Data operation Velocity (20 km/h)30m 40m

Mean 9.4637 11.3121Variance 1.8896 1.4886Maximum 6.6242 7.6275Minimum 12.7901 13.9346

and pitchmovement were neglected).The detailed data of thepreview distance was shown in Tables 2 and 3:

= (GazeOriginy + )GazeDirection.GazeDirection.

. (7)

is the vertical height of the world coordinate systemsorigin away from the ground; GazeOrigin. is the coordi-nate of the gaze origin; GazeDirection. is the coordinate ofthe unit vector along the gaze direction; GazeDirection. isthe coordinate of the unit vector along the gaze direction.

When the road curvature radius was 20m, the previewdistance increased slowly from 9,6253m to 10.1059m withthe adding of the velocity. When the velocity was constantly20 km/h, the drivers preview distance increased too whilethe road curvature radius increased from 30m to 40m.Noh et al. had used the preview time and the response ofneuromuscular system as main human factors to study thepreview distance at different velocities and different roadcurvatures whose conclusion was similar to ours [24].

The experiments result was consistent with the previoustests conclusion.With the consideration of the characteristicsof the preview distances distribution and change rule, thepreview distance-velocity regression analysis was done to getthe preview distance-velocity empirical equation (9). At thespeed of 20 km/h, according to the preview distance data with


the curvature radius of 30m and 40m, the preview distance-curvature radius empirical equation (10) was gotten by linerfitting. The fitting line is

= 0.025 + 9.358. (8)

is the preview distance; is velocity; is the correlationcoefficient, 2 = 0.98082, and the observed value =0.99036. = 5; with the significance level = 0.05, =0.99036 > 0.05 ( 2) = 0.878; the regression result wasobvious and the hypothesis the velocity change will affectthe drivers preview distancewas confirmed. To be consistentwith previous described parameters, the empirical equation( = 20m) was rewritten as follows:

= 0.025V + 9.358, = 20m, (9)

where V is velocity, km/h, is curvature radius, and is thepreview distance.

At the speed of 20Km/h, the empirical equation of thepreview distance-curvature radius was as follows:

= 0.185 + 1.529, V = 20 km/h. (10)

According to the relationship between the preview timeand the preview distance, = V, the empirical equation ofthe preview time was as follows:

= 0.09 +33.689

V, = 20m, (11)

= 0.0333 + 0.2752, V = 20 km/h. (12)

According to (11), the preview time calculated under thespeed of 10 km/h, 15 km/h, 20 km/h, 25 km/h, and 30km/hwas 3.489 s, 2.336 s, 1.774 s, 1.438 s, and 1.213 s, respectively;the average preview time is 1.70 s. According to (12), theaverage preview time is 1.44 s. Then (11) and (12) as empiricalequations of the preview time simulating the drivers previewbehavior were used to replace the fixed preview time in theAFS algorithm to confirm the location of the drivers previewpoint where the vehicle would arrive after the preview time.Then the headlamps deflection angle could be calculated.

4. The Simulation Analysis ofthe AFS Algorithm Simulating the DriversPreview Behavior

4.1. The Headlamps Light Distribution. The optical analysissoftware LucidShape was used to simulate the headlampslight distribution and acquire the headlamps isolux curveon the ground. In the following simulation process, theenvelope area of the isolux curve at a certain illuminationwas regarded as the headlamps irradiation area. Accordingto the relative relationship between the area and the road, thecontrol effect of the algorithm was verified by the headlampsirradiation effect. With the reference of ECE regulation, adipped headlight in accordance with the ECE regulation waschosen in our simulation and its optical structure was notchanged. The dipped headlights location settings were asfollows.

1030

80 70 60 50 40 30 20

10

0

10

0.250.3200.43050607080

0 50 100 15040

0

40

2

0.573

1.146

(m)

Figure 9: The simulation result of the headlamps illumination onthe road.

40

30

20

10

0

10

20

30

400 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

(m)

Figure 10: The envelope area of 321x isolux curve.

(1) The headlamps were installed in front of the vehicleand the headlamps were 1.2m apart and were sym-metrically on both sides of the vehicles longitudinalaxis.

(2) The installation height of headlamps was 0.65m.

The simulation result of the headlamps illumination onthe road was shown in Figure 9. The coordinate origin wasthe midpoint of the vehicles forefront. The horizontal andvertical coordinates were horizontal and vertical locationsrelative to the vehicle (unit m); different colors representeddifferent illuminations.

Meanwhile, because the roads curvature radius was rela-tively small in our simulation, to clearly verify our algorithmsirradiation effect, the envelope area of 321x isolux curve wasdrawn in LabVIEW as the headlamps irradiation area whichwas shown in Figure 10.

4.2. The Algorithm Verification. The vehicle model, drivermodel, and road model of CarSim were used in the simu-lation. The AFS algorithm and the real-time display of theheadlamps deflection angle, beam location, and irradiationeffect were realized in LabVIEW. By the cosimulation ofCarSim and LabVIEW, the bend driving simulating realenvironment was achieved. According to the calculatedheadlamps deflection angle, irradiation area, and the vehiclesfuture track, the algorithms control effect was verified.


1 2 3 4 5 6 7 8 9 10 11 12 13 14

Time (s)

Am

plitu

de (d

eg.)

The steering wheel angleThe proposed AFS deection angle

60.0040.0020.000.00

20.0040.0060.0080.00100.00120.00140.00160.00180.00

Figure 11:The headlamps deflection angle and steering wheel angle.

4.2.1. Simulation Conditions and Parameters. The vehiclesspeed was 30 km/h, the curvature radius of the bend was20m, and the preview time () was as follows:

= 0.09 +33.689

V= 1.213. (13)

4.2.2. The Simulation Result Analysis. The headlamps beamdeflection angle of our algorithm was shown in Figure 11.The dotted line represented the wheels steering angle and thesolid line stood for the headlamps deflection angle. It wasshown that the deflection angle calculated by our algorithmcould follow the wheels steering angle very well. In otherwords, the algorithm could respond to the change of vehiclestate quite well and controlled the headlamps deflectionangle. The control effect of the algorithm was shown inFigure 12. The dotted line represented the fixed headlampbeam.

It was shown in Figure 12 that the predicted track of ouralgorithmwas close to the roads centre line whichmeant thatthe track predictions accuracy was quite well. The vehicleslocation on the predicted track after the preview time wasused to guide the headlamps deflection. The headlamp wascontrolled to deflect to the direction of the bend and theheadlamps irradiation area covered most areas of the frontbend. When the vehicle was out of the bend, the algorithmcould make the headlamp beam back on the nondeflectionlocation to throw light on the front straight line.

4.3. The Contrast Analysis of the Algorithms Control Effect

4.3.1. The Simulation Conditions

Condition 1. The initial speed was 20 km/h and the driveraccelerated to 30 km/h in the time of 11 s13 s and kept thevelocity steady to pass through a half circle bend with theradius of 20m.The driving strategy was to make sure that the

vehicle has no lateral offset relative to the lanes center line.The time parameter rose to 1.774 s from 1.213 s as follows:

= 0.09 +33.689

V. (14)

Condition 2. The driver drove through two continuous quar-ter circles (the radiuses were 30m and 40m) at a constantspeed of 20 km/h. The driver model used in the simulationwas a model built-in CarSim. The time parameter rose to1.607 s from 1.274 s as follows:

= 0.0333 + 0.2752. (15)

The contrast algorithm was a nonpredictive algorithmproposed by Ishiguro and Yamada which was relativelymature [11]. The main point of this algorithm is to controlthe headlamps deflection to illuminate the location that thevehicle would arrive after 3 s. The contrast simulation resultwas as follows.

4.3.2. The Simulation Result Analysis

Condition 1. The headlamps deflection angle of the contrastalgorithm and our algorithm was shown in Figure 15. Theblack dotted line was the wheels steering angle curve, thesolid line was the headlamps deflection angle curve of ouralgorithm, and the dash dot linewas the headlamps deflectionangle curve of the contrast algorithm.

It was shown in Figure 13 that both algorithms headlampdeflection angle could follow the drivers steering angle.Relative to our algorithm, the contrast algorithms headlampdeflection angle was about 20 bigger at the moment 16 s.The control effect of these two algorithms was shown inFigure 14 by the headlamps irradiation area. The dottedline represented the fixed headlamp beam, the solid linerepresented the headlamp beam of our algorithm, and thelong dashed line represented the headlamp beam of thecontrast algorithm.

It was shown in Figure 14 that at the moment of =10.19 s, the vehicle just arrived at the bend; both the contrastalgorithm and our algorithm could predict the vehicles futuretrack well and make the headlamp deflect to the directionof the bend. Their deflection angles were basically the same.Most areas of the front bend were covered by the central zoneof the headlamps irradiation region which meant that theirillumination effects were basically the same.

At the moment of = 12.27 s, the vehicle was in the bend.The headlamps deflection angle of the contrast algorithmwasbigger than our algorithm which meant that its headlampbeam was more close to the inner side of the bend. Althoughthe headlamp had illuminated more regions of the frontbend, the control target of the contrast algorithm was toilluminate the location that the vehicle would arrive after 3 swhile the drivers preview time was 1.2131.774 s accordingto the empirical equation, so the region the driver caredfor was a region which was close to the vehicle (the purpleline in front of the vehicle) and it was not in the centre


5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

700 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

Y(m

)5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

The fixed illumination shapeThis paper illumination shape

5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

The fixed illumination shapeThis paper illumination shape

Figure 12: The front-lighting illumination area simulation schematic.

1 3 5 7 9 11 13 15 17 19 21

Time (s)

Am

plitu

de (d

eg.)

60.0040.0020.000.00

60.0040.0020.00

80.00100.00120.00140.00160.00180.00

The steering wheel angleThe contrast algorithm deection angleThe proposed AFS deection angle

Figure 13:The headlamps deflection angle and steering wheel angle(contrast algorithm and our algorithm diagram 1).

of the contrast algorithms illumination area which meantthat its illumination effect was not well. The future tracksprediction and irradiation region of our algorithmweremoreprecise.

At the moment of = 16.77 s, the vehicle was goingto be out of the bend. The headlamps deflection angle ofthe contrast algorithm was bigger and its illumination effectwas not ideal (relative to the fixed beam, its illuminationeffect could be worse). The future tracks prediction of ouralgorithmwasmore precise and the headlamps beam coveredmost of the front road which meant the illumination effectwas relatively well.

At the moment of = 18.86 s, the vehicle was out of thebend. Both the contrast algorithm and our algorithm couldquit from the bend illumination mode and control the head-lamp back to the nondeflection location to ensure the straight


10

5

0

5

10

15

20

25

30

35

40

45

500 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 9085

X (m)

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

10

5

0

5

10

15

20

25

30

35

40

45

50

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

10

5

0

5

10

15

20

25

30

35

40

45

50

Y(m

)

The fixed illumination shapeThis paper illumination shapeThe contrast illumination shape

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

X (m)

10

5

0

5

10

15

20

25

30

35

40

45

50

Y(m

)


t = 10.19 s t = 12.27 s

t = 16.77 s t = 18.86 s

90

9090

Figure 14: The front-lighting illumination area simulation schematic diagram 1.

4 6 8 10 12 14 16 18 20 22 24 26 28 30

Am

plitu

de (d

eg.)

60.00

40.00

20.00

0.00

60.00

40.00

20.00

80.00

100.00

The steering wheel angleThe contrast algorithm deection angleThe proposed AFS deection angle

Time (s)

Figure 15:The headlamps deflection angle and steering wheel angle(contrast algorithm and our algorithm diagram 2).

lines illumination whichmeant that their illumination effectswere the same.

Condition 2. The headlamps deflection angle of the contrastalgorithm and our algorithm was shown in Figure 15.

Both algorithms deflection angle could follow the inputof the drivers steering angle. Relative to our algorithm, thecontrast algorithms deflection angle was about 20 biggerwhen the steering wheel began to deflect.

The specified control effect was shown in Figure 16 bythe headlamps illumination area. The explanation of thecurves was similar to Figure 14 and the result was similar toCondition 1 which would not be analyzed again.

By the simulation result analysis of these two conditions,the conclusions were as follows.

(1) The future tracks prediction of our algorithm wasmore precise (the purple line in the image during theentire simulation was more close to the lanes centerline), so the algorithm would control the headlamps


5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

X (m)

5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Y(m

)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

X (m)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

X (m)

5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Y(m

)


0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

X (m)

5

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Y(m

)


t = 12.15 t = 15.01

t = 24.04 t = 26.17

Figure 16: The front-lighting illumination area simulation schematic diagram 2.

deflection more precisely to ensure the front roadsillumination effect well.

(2) Relative to the contrast algorithm, when the vehiclewas in the bend and was going to be out of the bend,the headlamps deflection angle of our algorithm wassmaller; the headlamps beam could cover the frontroad more well and make sure that the region thedriver was interested in was in the central zone ofthe headlamps beamwhichmeant better illuminationeffect.

(3) When the vehicle was just in the bend and was outof the bend after a while, the headlamps deflectionangle of both algorithms was the same and theirillumination effects were basically the same too.

In summary, the proposed algorithm could predict thevehicles future track more precisely; on the basis of it, theheadlamps deflection angle was controlled to improve thefront roads illumination condition effectively. Meanwhile,the proposed algorithm could make sure that the centralzone of the headlamps beam coveredmore regions which thedriver was interested in.

5. Conclusion

The previous road experiment has found that the previewdistance increased linearly with the velocity and the roadcurvature radiuss increasing. Based on this rule, the precisepreview distance was acquired by the eye tracker.The averagepreview time under the speed of 1030 km/h is 1.70 s. Under


the curvature radius of (2040)m, the average preview timeis 1.44 s. An empirical model of the preview distance was pro-posed which accorded with the drivers visual characteristics.

Based on the dynamic predict algorithm of the vehiclesfuture track using the drivers best preview accelerationmodel, the algorithms fixed preview time was modified byroad experiments and a full AFS algorithm was developedbased on the algorithm. The future tracks prediction of theproposed algorithm was precise.

Finally, by the cosimulation of CarSim and LabVIEW, theheadlamps illumination area and the vehicles future trackof different time were analyzed in detail in the bend drivingprocess. The illumination effect was analyzed in two aspectsand was compared with a mature nonpredictive algorithm.The result showed that the control effect was obvious and thealgorithm had a good application prospect.

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper.

Acknowledgments

This paper is supported by the National Natural Sci-ence Foundation of China (50975120), Specialized ResearchFund for the Doctoral Program of Higher Education(20120061110028), Jilin Provincial Research Foundation forTechnology Guide (20130413058GH), and Program forChang Jiang Scholars and Innovative Research Team inUniversity (no. IRT1017), China. Thanks are due to them.

References

[1] ECE Regulation No. 123, Uniform Provisions Concerning theApproval of Adaptive Front-lighting System (AFS) for MotorVehicles, 2007.

[2] L. Strambersky, M. Cejnek, and J. Martoch, Fast TrackStaticand Dynamic Bending Analysis, SAE Paper 2004-01-0440,2004.

[3] R. Huang, Automobile Adaptive Lighting System Development,Dalian University of Technology, 2008.

[4] L. Lifu and D. Qian, Research on the mathematical modelof bending mode of vehicle adaptive front-lighting system,Aotomobile Technology, no. 10, pp. 4346, 2010.

[5] X. Yu, Study on Automobile Adaptive Lighting Control SystemDalian, University of Technology, 2009.

[6] Z. Deng, The develop of AFS-adaptive frontlighting-system,China Light & Lighting, no. 6, pp. 811, 2006.

[7] S. McLaughlin, J. Hankey, C. A. Green, and M. Larsen, Targetdetection distances and driver performance with swivelingHIDheadlamps, SAE Paper 2004-01-2258, 2004.

[8] H. Rong, J. Gong, and J. Cao, Research on key technology ofAFS system auto electric parts, no. 5, pp. 1518, 2008.

[9] W. Pu, The profile and trends of Adaptive Front LightingSystem (AFS), Lamps and Lighting, no. 2, pp. 2224, 2009.

[10] F. A.-K. Ibrahim, Predictive Adaptive Front Lighting Algo-rithm for Branching RoadGeometry: United States, Tech. Rep.,US 7568822 B2, 2009.

[11] K. Ishiguro and Y. Yamada, Control technology for bendingmode AFS, SAE Paper 2004-01-0441, 2004.

[12] K.-H. Kim, D.-H. Yum, D.-K. Byeon, D.-Y. Kim, and D.-I. Lee,Improving drivers visual field using estimation of curvature,in International Conference on Control, Automation and Systems(ICCAS 10), pp. 728731, KINTEX, October 2010.

[13] M. F. Land, Predictable eye-head coordination during driving,Nature, vol. 359, no. 6393, pp. 318320, 1992.

[14] M. F. Land and D. N. Lee, Where we look when we steer,Nature, vol. 369, no. 6483, pp. 742744, 1994.

[15] M. Land and J. Horwood, Which parts of the road guidesteering? Nature, vol. 377, no. 6547, pp. 339340, 1995.

[16] M. Young, C. Chang, and Y. Chuang, A study of driversviewing field when driving on curvy roads, in Proceedingsof the the 3rd International Conference of Traffic & TransportPsychology, Nottingham, UK, 2004.

[17] H. Zhang and X. Zhang, Robust gain-scheduling gnergy-to-peak control of vehicle lateral dynamics stabilisation, VehicleSystem Dynamics, vol. 52, no. 3, pp. 309340, 2013.

[18] Z. Shuai and H. Zhang, Lateral motion control for four-wheel-independent-drive electric vehicles using optimal torqueallocation and dynamic message priority scheduling, ControlEngineering Practice, vol. 24, pp. 5556, 2014.

[19] Z. Shuai and H. Zhang, Combined AFS and DYC control offour-wheel-independent-drive electric vehicles over CAN net-work with time-varying delays, IEEE Transactions on VehicularTechnology, vol. 63, no. 2, pp. 591602, 2014.

[20] Z. Gao and F. Gao, Intelligent vhicle head lamp systemsimulating preview action of driver, CN 201010175695, China,2010.

[21] X. Guan, Z. Gao, and K. Guo, A hypothesis of steady previewand dynamic correction for driver model, Chinese AutomotiveEngineering, vol. 25, no. 3, pp. 1922, 2003.

[22] Z. Gao, X. Guan, Q. Li et al., Driver optimal preview longitu-dinal acceleration model, Automotive Engineering, vol. 24, no.5, pp. 434437, 2002.

[23] K. Guo,Vehicle Handling DynamicsTheory, Jiangsu Science andTechnology Publishing House, Nanjing, China, 2011.

[24] K. Noh, D. Jung, H. Choi et al., Development of ergonomicdriver model considering human factors, SAE Paper 2007-01-3584, 2007.

Submit your manuscripts athttp://www.hindawi.com

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttp://www.hindawi.com

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com

Volume 2014


Stochastic AnalysisInternational Journal of

Documents

939218