Automated Driving System based on Roadway and Traffic

Conditions Monitoring

Pavel Beresnev, Anton Tumasov, Dmitry Tyugin, Denis Zeziulin,

Valery Filatov and Dmitry Porubov

Nizhny Novgorod State Technical University n.a. R.E. Alekseev, Minina street, 24, Nizhniy Novgorod, Russian Federation

Keywords: Automated Driving System, Advanced Driver-Assistance Systems, Lane Detection, Traffic Sings Detection,

Obstacle Detection, Light Commercial Vehicles (LCV).

Abstract: In article development of a concept of advanced driver-assistance systems is considered. It consists of

several subsystems such as a warning about leaving the line, warning the driver about the possibility of

collision with an obstacle in the lane, detection of traffic signs. The concept of the system receives visual

information about the road scene and decides on the need for to correct the course and speed parameters of

the traffic. The component structure of system is shown. A number of methods are proposed (Hough

Transform, bird's eye view, etc.) to solve the task. Tests of a concept of advanced driver-assistance systems

are carried out. Based on the results of the tests, a technical task will be formulated for conducting

development work.

1 INTRODUCTION

For many years, studies on advanced driver-

assistance systems (ADAS) have been carried out to

reduce the number of accidents and improve safety

on the roads. The main algorithm of the advanced

driver-assistance systems while driving is the

collection of data from the sensors installed around

the vehicle, further analysis of the information,

detection of a critical situation on the road and

conducting measures to avoid or mitigate the

consequences

Currently the following ADAS are widely

distributed: the lane departure warning system,

obstacle detection system, traffic signs detection,

parking assistant, pedestrian detection system, night

vision system, adaptive cruise control, blind zone

monitoring system etc.

The developed systems are widely used by

automobile manufacturers. According to the

statistics from Bosch GmbH (The journal modern

electronics), the most common advanced driver-

assistance systems are: the lane departure warning

system (LDWS) and the advanced emergency

braking system (AEBS). Moreover, unintentional

departure from the occupied lane causes about 30%

of all accidents on the road.

The development of ADAS is carried out by

many large foreign automobile manufacturers.

However, there are no own developments and

solutions in the Russian Federation, which can be

installed on vehicles in series. To implement the

functions of ADAS on commercial vehicles adapted

to the operational conditions in Russian Federation,

the team of our university together with the GAZ

Group is working on the concept of the advanced

driver-assistance system on the GAZelle Next

vehicle with the possibility to correct the vehicle's

movement via the steering system.

At this stage the concept of the system consists

of:

– Lane departure warning subsystem, defined

by existing road markings;

– Driver warning subsystem, preventing the

possibility of collision with an obstacle on the

lane;

– Subsystem of traffic signs detection.

The concept of system should receive a stream

of data from the front sensors (optical and/or

infrared camera and thermal imager) for further

processing.

The concept of the system receives visual

information about the road scene and decides to

adjust the parameters of the direction and the speed.

Beresnev, P., Tumasov, A., Tyugin, D., Zeziulin, D., Filatov, V. and Porubov, D.

Automated Driving System based on Roadway and Trafﬁc Conditions Monitoring.

DOI: 10.5220/0006700303630370

In Proceedings of the 4th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2018), pages 363-370

ISBN: 978-989-758-293-6

363

2 ANALYSIS OF ADAS

At the first stage of the development, an overview of

information sources concerning the development of

advanced driver-assistance systems has been

performed.

In (Hoang, T., et al., 2016) a method for

detecting road lanes was proposed, which

distinguishes the dotted and solid lines of the road.

Based on the experimental results with the Caltech

open database, this method showed excellent results

over traditional methods for detection road lanes.

The paper (Shanti B., and Ganorkar S., 2015.)

shows the developing of an algorithm for a lane

departure warning system. The system can be used

successfully to reduce the number of road accidents

that have occurred due to the sleepiness or

inexperience of the driver. This paper shows the

research and comparison of the available algorithms.

Authors developed an algorithm for the image

processing, which uses one camera installed on the

vehicle. The stability of the algorithm has been

tested in different conditions, such as lightning,

shadows from adjoining building, vehicles etc. The

algorithm is implemented in two stages. At the first

stage, the algorithm finds the point of descent on the

road, and then carefully selects the segments. The

lines of the roadway were made by using

information about the point of the road boundary. A

heuristic filter was used to check and detect the road

lanes. The Kalman's filter is used in order to track

the road lane and processing the critical sections

where the detection algorithm does not work.

The paper (Janda F., et al., 2013) is presenting a

reliable real-time approach for determining the

edges of roads by obtaining information from the

radar and video camera. The road edge is defined as

the transition from asphalt to off-road. The

integration of the multi-lane detection system is

shown, that makes this approach independent of the

number of lanes and the visibility of the lane marks.

System's performance is evaluated by using the

reference data. Information about road geometry,

curvature and the relative position of the vehicle is

critical for advanced driver-assistance systems

(ADAS). The detection of the road edge is the main

component for proposed functions, such as

preventing the crossing road edge, which keeps the

vehicle on the under control area. This work is the

part of the European project «interactiveIVe» that

explores the new technologies and approaches to

improving vehicle's safety via an integrated

platform.

Paper (Pydipogu P., et al., 2013) describes the

application of the IPM (Inverse Perspective

Mapping) method. The system receives a video

image by the camera installed on the vehicle, and

then uses several processes to detect objects and

road markings. A universal methodology was used

to detect lanes and objects. A simple heuristic

method was developed and it gives reliable results of

detecting and tracking objects and lanes on a video

stream. The heuristic method also gives effective

results in the detection and tracking of several

vehicles, regardless of distance.

An obstacle detection method based on moving

cameras was developed in (Shah V., et al., 2016).

This method was used to detect various obstacles

(animals, traffic signs, obstacles, roadway

irregularities), knowing the size of the roads. A new

technique for detecting obstacles through moving

cameras was proposed, that has several limitations in

comparison with stationary cameras. The latest

research trends were analyzed in this paper.

In the paper (Sumi K., and Arun Kumar M.,

2017.) various methods of detection and recognition

of traffic signs are explained. To detect the traffic

signs and recognize their text, different methods are

applied in order to get the desired accuracy. A

comparative research of the methods is carried out

and their effectiveness is explained. The authors of

the paper suggest dividing the problem of detecting

traffic signs into two stages. The first step is to find

out the area of the desired traffic sign. The second

stage is character recognition.

The system in (Fifik M., et al., 2010) is based on

a two-stage traffic signs detection scheme. Input is

the image that contains the information for

extraction. A block called "Hypothesis Generation"

(HG) generates promising hypotheses in the form of

a Region of Interest (ROI). After this, the part called

Hypothesis Verification (HV) checks or rejects the

previously stated hypothesis. As the output, we have

an image with ROI, which is recognized or

determined during subsequent processing. This

system can be divided into two subsystems. One

subsystem is "fast", consisting of a CSR block and a

classification block; the second is a "slow"

subsystem consisting of image segmentation blocks,

an invariant allocation function, functional memory,

function modification, and type classification blocks.

Also, experiments with this system were carried out.

The average recognition speed of the system was

86% for the Hue, Saturation, Value (HSV) color

area and 89% for the red, green, blue (RGB) color

area. The best results achieved were 3-4 seconds for

image processing.

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

364

Benchmarking of existing solutions in the area of

advanced driver-assistance systems and autopilot

motion was performed. As the comparative criteria a

list of functions were defined:

1) Means of obtaining information;

2) Detection of the road marking;

3) Detection partially erased road marking;

4) Detection of the road edge;

5) System «follow me»;

6) Detection of vehicles moving towards;

7) Lane assist;

8) Night Vision System;

9) Movement in bad weather and road

conditions (rain, snow, etc.);

10) Algorithms and method of functioning;

The objects of benchmarking were the world

leaders of the development in the area of ADAS and

unmanned ground vehicles. These are General

Motors, Google, nVidia, Tesla, Volkswagen,

Mercedes Benz, Baidu and others.

Regarding the first point of benchmarking almost

all the manufacturers use cameras, it is also possible

to use LIDARs and radars.

Vehicles of all companies are able to detect the

road marking. But only General motors, Tesla and

Volkswagen can detect partly erased road marking.

Only General Motors, Google, nVidia, Baidu can

detect the edge of the road.

To the fifth point of benchmarking list

correspond the vehicles by Google, nVidia, Tesla,

Volkswagen and Baidu. Also these vehicles are able

to detect other vehicles moving towards.

The lane assist is realized practically at all

companies, the big development of this system is

realized at large automobile manufacturers, such as

Volkswagen, Tesla and Mercedes Benz.

Advanced driver-assistance systems in all

presented above vehicles are functioning at night.

But in such bad road conditions as snow, rain, fog,

best work performance is confirmed only by Google,

nVidia, Tesla and Volkswagen.

By having considered information sources and

carrying out benchmarking of existing solutions in

the area of development of advanced driver-

assistance systems, it was revealed that:

– At present, interest in developing or

improving approaches to identifying

obstacles, road lanes and signs is increasing

(as evidenced by the large quantity of

different articles in foreign sources).

– The use of cameras as a means of solving

tasks is topical along with the growth of

appropriate technologies.

– The use of cameras is the cheapest method of

solving the problem of detection objects,

lanes and road signs.

– Based on the conducted benchmarking, it can

be concluded that the integration of advanced

driver-assistance systems into vehicles of the

largest automotive manufacturers has

increased and large-scale research in this area

of leading companies (Google, nVidia, Tesla,

Volkswagen, Mercedes Benz, etc.) are

carried out.

3 DEVELOPMENT OF ADAS FOR

LCV

3.1 Composition of ADAS

Based on the review of actual literature and

benchmarking of advanced driver-assistance

systems, the components of the concept of the lane

detection system, the traffic signs detection and the

obstacles detection on the road was determined.

As optoelectronic sensors, the Basler ace

acA1300-200uc cameras are used. The location of

the cameras is in the sides and in the middle of the

windscreen. The video stream from the cameras is

processed using the nVidia Jetson TX2 platform that

has considerable computing power and small size

that significantly affects the compactness of the

entire system. Based on the analysis of driver

assistance systems, video cameras are not sufficient

for reliable and efficient operation of the entire

recognition system. It is necessary to duplicate

systems of recognition of the road scene by sensors

that can function in difficult operating conditions,

such as erased road lane, poor illumination of roads,

presence of animals on the roadway, etc. In order to

solve the above problems, it is proposed to use

thermal imaging sensors in the system concept. The

TITAN IP thermal imager is used in our system. A

wide range of temperature exploitation allows it to

function in snow, rain, fog and to recognize not only

dynamic objects of the road scene (cars, pedestrians,

animals), but also to determine the boundaries of the

roadway and road markings. The structural scheme

of interaction of the concept of the advanced driver-

assistance systems is presented in Figure 1.

Automated Driving System based on Roadway and Trafﬁc Conditions Monitoring

365

Figure 1: Structural scheme of the interaction of the

driver-assistance systems.

3.2 Detection Method

There are many different approaches and methods

for recognizing road lanes and road signs: B-spline

model (Li W., et al., 2014, Deng J., et al., 2013);

Hyperbola-pair lane model (Tan H., et al., 2014);

Lane geometrical model (Zhou S., et al., 2010);

Vehicle directional control model (DIRCON)

(Litkouhi B., et al., 1993); Quadratic function

model (Yoo H., et al., 2013); IPM model (Shin J., et

al., 2014, Lu W., et al., 2014); Linear or parabolic

model (Mu C., and Ma X., 2014.) etc.

Based on the results of the analysis of possible

recognition methods, as well as taking into account

the efficiency of computing power, two methods of

recognizing road lanes and signs were chosen, this is

the Haar Feature and the Hough Transform (Máthé

K., and Buşoniu L., 2015). The use of these methods

for solving the problem is compared.

Let us single out a number of criteria and

compare the proposed methods.

Basis of the Method:

1. Haar Feature. The Viola-Jones method,

wavelet, Haar primitives;

2. Hough Transform. In the simplest case, the

Hough transform is a linear transformation to detect

straight lines or boundaries.

Training:

1. Haar Feature. A database of images of the

object sought in real-world conditions and a

database of images without the desired object are

needed. The cascade learns by examples, creating a

set of rules for finding this object in xml format;

2. Hough Transform. No.

Identification Method

1. Haar Feature. In the entire matrix of the

frame, it searches for a region suitable for a

particular xml rule;

2. Hough Transform. The Hough's Transform

algorithm uses an array called the accumulator to

determine the presence of a direct (special case) y =

mx + b. The dimension of the accumulator is equal

to the number of unknown parameters of the

Hough's area. Search is carried out by means of

mathematical transformations.

Restrictions

1. Haar Feature. It is necessary to collect raw

data directly at the target location of the detector.

Any change in incoming data (changes in

illumination, weather conditions, color

characteristics, camera angle) adversely affects the

result of detection. The algorithm is sensitive to the

rotation angle of the sought object;

2. Hough Transform. The efficiency of the

algorithm follows from the quality of the input data:

the boundaries of the figures during the

preprocessing of the image must be clearly defined.

The use of Hough's transform on noisy images is

difficult. For noisy images, a preprocessing step is

required to suppress noise.

Advantage

1. Haar Feature. High speed of work, the ability

to detect complex objects;

2. Hough Transform. High speed of work. It is

possibility the detection of simple geometric shapes.

Based on the analysis, the Hough Transform was

chosen as the main algorithm for detecting traffic

signs and lanes.

3.2.1 Obstacles Detection on the Road

The TenforFlow Object Detection API is used, that

allows creating and training object detection models

based on neural networks to determine objects on

the route,

The TensorFlow Object Detection API is an

open source platform based on the TensorFlow

library that allows you to create and training object

detection models. The model is a set of files that

contain information necessary for the recognition of

certain objects. The finished model is the result of

training of the neural network. Before training, a

description file for detection objects is created, as

well as set of data for each object on which the

neural network will conduct it’s training. The

training and debugging of neural network requires

large computing power.

In the first stage (Figure 2), the obstacle

detection system was tested on a video stream from

a dash camera with a minimum number of obstacles

(vehicles).

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

366

Figure 2: The first tests of the system on video files.

At the end of the first stage, the system was

modernized and tested on a real road conditions.

Figure 3 shows the frames tests of the system.

Figure 3: Example of system tests on a video stream of

road conditions in the city of Nizhny Novgorod.

Upon completion of the system testing, a

conclusion was made about the operability of this

method.

3.2.2 Traffic Sign Detection

The system of automatic recognition of traffic signs

is an important part of ADAS. It is designed to

notify the driver of the presence of traffic signs on

the road. The system can help the driver follow the

speed limit established on the road section, observe

travel restrictions, overtaking, etc.

Based on the chosen Hough Transform, the

model of recognizing road signs on the roadway was

developed. The development of the model was

carried out with the open source library OpenCV

(OpenCV: library) of computer vision algorithms,

image processing and numerical algorithms for

general purpose.

In order to design the algorithm, a number of

traffic signs were used in accordance with the

national standard of the Russian Federation

№52290-2004 (National standard of the Russian

federation). The list of traffic signs is presented in

Table 1.

Table 1: List of used road signs.

Crosswalk The main road

No overtaiking Speed limit

This algorithm allows finding the necessary sign,

only if it is available in the database.

For the operation of this algorithm, an image

conversion circuit has been designed.

Figure 4: The image conversion scheme.

Based on this conversion scheme, a number of

tests were carried out. Screen shots of the results are

shown in Figure 5.

Figure 5: Screenshot of the system's output.

With regard to the recognition of road signs, the

system showed an efficiency that meets the

requirements, which is 96%, but it is necessary to

expand the list of recognizable road signs for the

most complete analysis of the road scene.

Automated Driving System based on Roadway and Trafﬁc Conditions Monitoring

367

3.2.3 Lane Detection

The software of the road line detection system is

based on the bird's eye view algorithm. Using this

algorithm, with respect to solving the current task,

allows you to display the selected section of the

roadway in a distorted perspective. The scheme of

this algorithm is shown in Figure 6.

As a region of interest (ROI), the trapezoidal

image area is selected. This approach allows

excluding unnecessary artifacts (untreated frame

substrate) after a perspective distortion in the case of

using a rectangular shape as the ROI.

By selecting the optimal parameters, an

acceptable perspective view of the road section can

be obtained. Then, there is a transformation of the

image of ROI to the distorted perspective image.

To further process the resulting mapping, the

matrix must be transferred from the RGB color

space to the HSV color space. This space is less

sensitive to light and shade changes in the frame,

which allows increasing the stability of the

algorithm on complex sections of the road.

Figure 6: The scheme of the bird's eye view algorithm.

After the transition to the desired color space, the

important preprocessing of image is binarization.

The purpose of binarization is to exclude from

the frame all insignificant pixels that are not related

to road markings (all pixels outside the color range

of white and yellow markings).

To search for lines in the frame, the above-

mentioned Hough transform is used.

When the function receives the binarized image

of the prospective road as the input, it returns the

vector of lines that can be used to identify the lines

found in the frame.

After receiving a vector of lines, it is necessary

to approximate all set in order to get one line

corresponding to the position of the road line in the

frame.

After carrying out virtual tests and optimizing

the system for detecting road marking, an algorithm

was further developed for warning about lane

departure and tests were conducted in real road

conditions.

The algorithm of the lane departure warning

system is as follows. There are 3 warning zones:

green - the car is in the center of the lane; yellow -

deviation from the center of the lane is in the range

from 0.2 to 0.3 m; red - deviation from the center of

the lane is more than 0.3 meters (Figure 7-9).

During the tests, the car speed was about 70 ± 10

km/h, entering the turn by a radius of curvature of at

least 100 m and keeping the speed at a given level.

After that the car performed a successive smooth

approach to the road marking at a speed of 0.3 m/s.

About 100 tests have been performed.

Figure 7: Example of the warning algorithm. Green zone.

Figure 8: Example of the warning algorithm. Yellow zone.

Figure 9: Example of the warning algorithm. Red zone.

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

368

The triggering did not happen, if in the process

of making the vehicle closer to less than 30

centimeters with the line marking, no alert was

given.

Based on the results of the tests, the following

data were evaluated: the number of correct alarms

when approaching the marking lines; the number of

false alarms when approaching; number of failures

when approaching.

The calculated accuracy of the recognition of

marking lines was 0.76. The completeness, which

shows the number of correct responses from the total

number of real intersections of the road marking

line, was 0.92. The indicator of the quality of the

system was 84%.

3.3 Steering System

The developed concept of the system is planned to

be installed on the GAZ Group's light commercial

vehicles (GAZ Group).

GAZelle NEXT vehicles are equipped with a

rack-and-pinion with a hydraulic booster steering

system, which causes difficulties for the integration

of advanced driver-assistance systems due to the

cumbersomeness of the whole system and the

complex technique of controlling the pump of the

hydraulic system. Thus it was suggested to use a

steering rack with an electric motor with the control

capability. This steering rack shows similar strengths

on the steering wheel that allows it to be used for

GAZ Group commercial vehicles.

Based on the received and processed data from

the lane detection system, the nVidia Jetson TX2

platform sends a signal to the steering controller

with the electric motor via the CAN bus of the

vehicle.

Figure 10: Steering system.

The controller controls the electric motor of the

steering rack, sending the necessary signals via the

data bus. This method is the most secure and allows

minimizing the possibility of losing control of the

steering rack. The scheme of the system is shown in

Figure 10.

Knowing the necessary distance to the side lane,

the width of the lane and the width of the vehicle,

the system is able to correct the movement of the

vehicle, keeping it on the lane and does not allow to

leave the lane.

4 CONCLUSIONS AND FUTURE

WORK

Information sources were reviewed and

benchmarking of existing solutions in the field of

driver assistance systems, including autopilot

technologies have been made.

Based on the analysis, the component

composition of the concept system and the

interaction scheme have been proposed.

Methods for solving the problem are proposed.

The efficiency of the proposed methods was

confirmed by a number of tests.

Combining systems for detecting road markings,

signs, and obstacles along the way provides the

opportunity to obtain the necessary information

about the road scene in order to implement correct

control actions on the steering mechanism.

The scheme of course orientation control by

means of electromechanical system is developed.

Timely impact of the electric motor through the

servo-drive to the steering rack allows keeping the

car in the lane, thereby increasing safety and

reducing the probability of unintentional crossing of

the lane.

At this stage of the project the structure of

system of the ADAS-3 (SAE J3016) level is

developed.

The problems in the technical implementation of

the active steering system for light commercial

vehicles of the GAZ Group have been solved.

The system works in various weather conditions

and the heavy climate of the Russian Federation.

The next stage of the project is the development

of design documentation for various vehicle

systems, software and data transmission networks

that integrate all subsystems into a single network in

the on-board computer.

The prototype of the system concept will be

installed on the GAZelle Next vehicle and examined

at the GAZ Group’s test site for debugging all of the

components. Based on the results of the tests,

Automated Driving System based on Roadway and Trafﬁc Conditions Monitoring

369

recommendations will be given in order to to

compile the Technical Assignment for development

work.

ACKNOWLEDGEMENTS

This research done with the financial support from

Ministry of Education and Science of the Russian

Federation in the frame of the complex project “The

establishment of the high-tech manufacturing of safe

and export-oriented GAZ vehicles with autonomous

control systems and the possibility of integration

with the electric platform on the base of components

of Russian production” under the contract

№03.G25.31.0270 from 29.05.2017 (Governmental

Regulation №218 from 09.04.2010).

REFERENCES

Deng J., Kim J., Sin H., Han Y., 2013. Fast lane detection

based on the B-spline fitting. Int. J. Res. Eng. Tech., 2,

134–137.

Fifik M., Turán J., Ovseník L., Fazekas K., 2010.

Experiments with a Transform based Traffic Sign

Recognition System, Proc. of 17th International

Conference on Systems Signals and Image Processing

IWSSI P2010, pp. 227-230, June 17–19.

GAZ Group. Retrieved October 30, 2017, from:

http://gazgroup.ru/

Hoang, T., Hong, H., Vokhidov, H., Park, K., 2016. Road

Lane Detection by Discriminating Dashed and Solid

Road Lanes Using a Visible Light Camera Sensor.

Sensors 2016, 16, 1313.

Janda F., Pangerl S., Schindler A., 2013. A Road Edge

Detection Approach for Marked and Unmarked Lanes

Based on Video and Radar, 16th International

Conference on Information Fusion, July 9-12.

Litkouhi B., Lee A., Craig D., 1993. Estimator and

Controller Design for Lanetrak, a Vision-Based

Automatic Vehicle Steering System. In Proceedings of

the 32nd Conference on Decision and Control, San

Antonio, TX, USA, 15–17; pp. 1868–1873.

Li W., Gong X., Wang Y., Liu P., 2014. A Lane Marking

Detection and Tracking Algorithm Based on Sub-

Regions. In Proceedings of the International

Conference on Informative and Cybernetics for

Computational Social Systems, Qingdao, China, 9–10;

pp. 68–73.

Lu W., Rodriguez F., Seignez E., Reynaud R., 2014.

Monocular Multi-Kernel Based Lane Marking

Detection. In Proceedings of the 4th Annual

International Conference on Cyber Technology in

Automation, Control, and Intelligent Systems, Hong

Kong, China, 4–7; pp. 123–128.

Máthé K.; Buşoniu L., 2015. Vision and Control for

UAVs: A Survey of General Methods and of

Inexpensive Platforms for Infrastructure Inspection.

Sensors 2015, 15, 14887-14916.

Mu C., Ma X., 2014. Lane detection based on object

segmentation and piecewise fitting. TELKOMNIKA

Indones. J. Electr. Eng., 12, 3491–3500.

NATIONAL STANDARD OF THE RUSSIAN

FEDERATION №52290-2004 Traffic control devices.

Traffic signs. General technical requirements.

On-road Automated Vehicle Standards Committee, 2014.

SAE J3016: Taxonomy and Definitions for Terms

Related to On-Road Motor Vehicle Automated

Driving Systems. SAE International.

OpenCV library. Retrieved October 30, 2017, from:

https://opencv.org/

Pydipogu P., Fahim M., Shafique M., 2013 Robust lane

detection and object tracking In relation to the

intelligence transport system. Master’s Thesis

Electrical Engineering Signal Processing.

Shah V., Maru S., Jhaveri R., 2016. An Obstacle Detection

Scheme for Vehicles in an Intelligent Transportation

System, Int. J. Comput. Netw. Inf. Secur., vol. 8, no.

10, pp. 23–28.

Shanti B., Ganorkar S., 2015. Real-Time Lane Detection

for Driving System Using Image Processing.

International Research Journal of Engineering and

Technology (IRJET), Volume: 02 Issue.

Shin J., Lee E., Kwon K., Lee S., 2014. Lane Detection

Algorithm Based on Top-View Image Using Random

Sample Consensus Algorithm and Curve Road Model.

In Proceedings of the 6th International Conference on

Ubiquitous and Future Networks, Shanghai, China, 8–

11; pp. 1–2.

Sumi K., Arun Kumar M., 2017. Detection and

Recognition of Road Traffic Signs - A Survey.

International Journal of Computer Applications (0975

- 8887), Volume 160 - No.3.

Tan H., Zhou Y., Zhu Y., Yao D., Li K., 2014. A Novel

Curve Lane Detection Based on Improved River Flow

and RANSA. In Proceedings of the International

Conference on Intelligent Transportation Systems,

Qingdao, China, 8–11; pp. 133–138.

The journal modern electronics. Retrieved October 30,

2017, from: https://www.soel.ru/rubrikator/rynok/

Yoo H., Yang U., Sohn K., 2013. Gradient-enhancing

conversion for illumination-robust lane detection.

IEEE Trans. Intell. Transp. Syst., 14, 1083–1094.

Zhou S., Jiang Y., Xi J., Gong J., Xiong G., Chen H.,

2010. A Novel Lane Detection Based on Geometrical

Model and Gabor Filter. In Proceedings of the IEEE

Intelligent Vehicles Symposium, San Diego, CA,

USA, 21–24; pp. 59–64.

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

370