Emergency Corridor Building on Multi-Lane Motorways with

Autonomous Model Cars

Jurij Kuzmic, Günter Rudolph and Fabian Ostermann

Department of Computer Science, TU Dortmund University, Otto-Hahn-Str. 14, Dortmund, Germany

Keywords: Emergency Corridor Building, Model Cars, Standing Traffic, Lane Detection, Filtered Canny Edge Detector,

Autonomous Driving, Vehicle-to-Vehicle (V2V) Communication, Autonomous Traffic.

Abstract: This paper introduces an algorithm for forming an emergency corridor on motorways with autonomous

vehicles. This algorithm can be used in slow-moving traffic and in standing traffic scenarios. In addition,

several autonomous model vehicles were assembled for the experiments in this work in order to test this

algorithm in real-world use. Furthermore, a model motorway was constructed that resembles a real three-lane

motorway. The Filtered Canny Edge Detector algorithm, also previously published by us, is used to recognise

the lanes of the motorway from camera images. Also, in this work, this lane detector is further extended and

improved for use with model vehicles in the real environment. Our experiments also show successful

emergency lane formations for four different events. Finally, possible future work in this area is presented.

1 INTRODUCTION

Unforeseeable accidents happen again and again on

roads and motorways. Sometimes serious accidents

can occur on the motorways as vehicles are travelling

at high speeds. This leads to traffic jams and road

closures. In those cases, most countries have

regulations for forming an emergency corridor (or

rescue lane) for emergency vehicles arriving at the

scene of an accident. But human drivers sometimes

forget to form such an emergency lane for police and

rescue vehicles when in a traffic jam. In extreme

cases, where the traffic comes to a hold in a chaotic

constellation due to a sudden accident, it is no longer

a straightforward task to form the rescue corridor.

When the vehicles come to a standstill, they are

usually too close to each other. In addition, there is

the problem that even if a rescue lane can be formed

after the vehicles have stopped and one emergency

vehicle drives past, the drivers of the standing

vehicles close the rescue lane again. This repeatedly

leads to the obstruction of other police or rescue

vehicles.

The problem is not new and has been discussed

for some time (Dębiński, Jukowski and

Bohatkiewicz, 2018). Currently, there is no optimal

solution. Some countries impose heavy fines for

obstructing emergency vehicles (e.g., Austrian

Federal Ministry of Mobility, 2023). To counteract

this problem, an Emergency Vehicle Warning System

has already been introduced, which uses radio

communication to warn other vehicles about an

emergency vehicle (Buchenscheit, Schaub, Kargl and

Weber, 2010). To completely eliminate this problem,

the emergency corridor must be formed automatically

by the autonomous vehicles that probably soon are

likely to be approved for regular road traffic. The

question is not if, but when autonomous vehicles will

come onto our roads.

According to our research, at the time of the

development of our algorithm, there was only the

work we published in the field of automatic formation

of an emergency corridor with autonomous vehicles

(Kuzmic and Rudolph, 2020). In addition, there were

only three patents for warning and communication

systems for the formation of the emergency lane

through vehicle-to-vehicle communication (Colella

and Herman, 2018; Marin, 1998; Siegel, 2003). This

allows for sharing information between road users,

for example about accident location or the route of an

approaching emergency vehicle.

To reduce the required budget for our study to a

minimum, we are using model vehicles. Of course,

the use of real cars would be preferable, but since we

would need at least a dozen of them to simulate a

rescue corridor situation, this is not achievable even

for highly equipped research labs. With our approach,

Kuzmic, J., Rudolph, G. and Ostermann, F.

Emergency Corridor Building on Multi-Lane Motorways with Autonomous Model Cars.

DOI: 10.5220/0012731700003705

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 9th International Conference on Internet of Things, Big Data and Security (IoTBDS 2024), pages 321-328

ISBN: 978-989-758-699-6; ISSN: 2184-4976

321

scaling future experiments to having tens or hundreds

of cars is in a reasonable scope. In contrast to a

computer simulation, the model cars experience

problems of real world scenarios, for example, sensor

noise, inaccuracies in motor control, or blurred

images obtained from a camera in motion.

2 ALGORITHM FOR

EMERGENCY CORRIDOR

BUILDING

Algorithm 1 is developed for the formation of the

emergency corridor in standing traffic (when the

traffic has come to a standstill). The algorithm works

with four states. The normal state is called

NORMAL_FORWARD. This state’s logic was already

published in (Kuzmic and Rudolph, 2020) and

handles the slow-moving traffic scenario. Vehicles

usually are driving at high speeds on the motorway.

If the speed falls below a certain value (e.g. 30 km/h)

because of heavy traffic, the vehicles begin to form a

rescue corridor as they drive. This means that the

vehicles on the last lane (the leftmost lane, as we

assume right-hand traffic) move to the left lane

marking and the vehicles on the adjacent lane move

to the right lane marking. The vehicles in the other

lanes do not change their lane position. Counting of

the lanes starts in the direction of driving from right

to left, as the first lane is decided to be the lane one

drives onto first. Please note, that our algorithm also

works for left-hand traffic.

For unforeseeable reasons, it can happen that

vehicles come to a sudden standstill and human

drivers forget to form an emergency corridor. The

rescue lane for police and rescue vehicles is then

missing. With our algorithm, the drivers of these

vehicles can attempt to safely form an emergency

corridor out of such a chaotic situation at the push of

a button, or in the case of fully automated vehicles,

the cars decide to do this themselves.

Algorithm 1, requires the generation of a local

map for each individual vehicle. This means that all

vehicles share information about their positions,

including their current lane number. By sending the

information of their own identification number (in

practice, this can be the licence plate number), their

own lane position and the identification number of the

vehicle in front, it is possible to calculate the order of

the vehicles on all the lanes and to create a local map.

Then, the vehicles can decide which vehicle has to

start with forming the emergency corridor. Three

further states are used to form the corridor from

standing traffic. They allow to form the rescue lane

systematically (row by row). The vehicles closest to

the vehicles involved in the accident will start. As

soon as a vehicle has completed its task, it notifies the

vehicle behind it on the same lane. The corresponding

states WAIT, EXEC_BACKWARD, and

EXEC_FORWARD will be explained in more detail

in Section 4.

Algorithm 1: Formation of an emergency corridor in a

standing traffic.

3 AUTONOMOUS MODEL CARS

The SunFounder PiCar-S Smart Car kit for the

Raspberry Pi (SunFounder, 2022) was used to

assemble the model vehicles from the model making

area. This already contains several motors, wheels,

control plates and sensors for the construction (Fig. 1,

left). The cost of each individual kit was around 90

Euros at the time of development (the Raspberry Pi

not included). Our own measurements show that the

model vehicles reach speeds of 0.61 m/s (2.22 km/h).

IoTBDS 2024 - 9th International Conference on Internet of Things, Big Data and Security

322

We have adapted and expanded the SunFounder

PiCar-S Smart Car kit for our use case. Figure 1

shows a comparison between the original model

vehicle and our modified version.

Figure 1: SunFounder PiCar-S Smart Car kit for Raspberry

Pi. Left: Original. Right: Modified.

3.1 Construction of the Model Cars

The body of the model vehicles consists of several

die-cut acrylic panels, which are fixed in place with a

few screws. This results in slight inaccuracies in the

steering of the model vehicles. A stepper motor is

responsible for steering the model vehicles. But, the

steering can only be carried out in steps of 2° instead

of 1°: If the stepper motor is moved in 1° steps, the

front wheels do not move. This is due to inaccuracies

in the motor control design.

The next step was to add a Raspberry Pi 3 B

(Kofler, Kühnast and Scherbeck, 2021) and a

Raspberry Pi V2 camera module (RaspberryPi, 2016)

to the kit. The camera module allows to optically

detect the lane markings. The camera was installed in

a suitable camera housing. This housing was attached

with two spacers and fixed at a certain height from the

motorway at a certain angle to the motorway. The

ultrasonic sensor for distance measurement was

installed lower so as not to restrict the camera's view.

Figure 2: Extensions of the SunFounder PiCar-S Smart Car

kit for Raspberry Pi. Left: QR-Code bumper. Right:

Distance measurement with ultrasonic sensor.

A black cardboard bumper was attached to the back

of each model vehicle so that the ultrasonic sensors

find a straight surface for measuring the distance to

the model vehicle in front (Fig. 2). The affixed QR

code provides the information of the vehicle's

identification number (ID).

3.2 Construction of the Model

Motorway

A miniature model motorway was set up to carry out

the experiments. This allowed to test the previously

developed algorithm in different situations. In

addition, this test track could be used to test the

improved lane detection with the model vehicles and

to adapt it for this purpose (details follow in Section

3.3). The test motorway was also adapted to the size

of the model vehicles used. The test track resembles

a three-lane motorway (Fig. 3, left). To carry out the

planned experiments also on a two-lane motorway,

one of the lanes can be “closed”. Eight metres of

impact sound insulation for laminate was used for the

ground of the track. This material resembles the

colouring of a real road surface. White insulating tape

was used to apply the lane markings. Figure 3 (right)

shows the test model motorway including the nine

modified model vehicles that were available for our

experiments.

Figure 3: Construction of the three-lane model motorway.

Left: Motorway without model cars. Right: Motorway with

model cars.

3.3 Lane Detection

For lane detection with the model vehicles, the

previously published Filtered Canny Edge Detector

algorithm (Kuzmic and Rudolph, 2021) was used. It

is further extended and optimized for the model

vehicles in the present paper. All that is written in

bold font is a novel addition, that substantially

improves lane detection on real camera images.

0. As a preliminary work, the camera should be

aligned and calibrated. Only the vital areas

should be visible in the image (only the

road). E.g., the bonnet of the car, low-

Emergency Corridor Building on Multi-Lane Motorways with Autonomous Model Cars

323

hanging power lines, etc., should not be

visible.

1. Create a greyscale image from the coloured

input image.

2. Create a Canny edge image from the

grayscale image (Canny, 1986).

3. Apply probabilistic Hough transform

(Stephens, 1990) to the edge image

(2D coordinates for P and Q).

4. Calculate the gradient m for each straight

line.

5. Calculate the angle with atan(m) from the

gradient m.

6. Ignore all lines with an angle θ below α=25°

and above β = 80° (α < θ < β).

7. Calculate the points of intersection with the

X-axis IX for the remaining lines.

8. Sort left and right lines from the centre of

the image B into separate data fields.

9. Ignore left lines with a negative gradient

m<0 and right lines with a positive

gradient m>0.

10. Save intersection points SX, angle θ,

gradient m and P and Q coordinates of

the lines for left and right lines

respectively.

11. Calculate intersection points of left and

right lines. Accept them in a predefined

area between line intersection point 1 and

line intersection point 2.

12. Origin of these lines must be in the lower

two-thirds of the image.

13. From the centre of the image B, find the

smallest distance D = |IX-B| left and right.

14. Search in a predefined radius R for further

intersection points IX left and right.

15. Lines found in the predefined radius R

must contain a similar angle to the X-axis.

Angle threshold W with |θ

- θ

| < W.

16. Calculate an average function (straight line)

left and right from the found lines.

17. Calculate the centre of the track from these

two functions left and right.

The runtime measurements were carried out on nine

different autonomous model vehicles. By

implementing and executing the Filtered Canny Edge

Detector algorithm with multithreading on the

Raspberry Pi 3 B, we achieved slightly below 30

frames per second (FPS) as shown in Table 1. Also,

Table 1 shows that the model vehicle with ID 9

achieves only 24.5 FPS. This is due to an individual

hardware defect of this particular model car.

Table 1: Runtime of lane detection on the Raspberry Pi 3 B.

Model Car [ID] Lane [No] Runtime [FPS]

3 29.69

2 29.33

1 29.24

3 29.46

2 28.88

1 28.00

3 29.44

2 29.14

1 24.50

Figure 4 shows the improved Filtered Canny Edge

Detector in use on real-world camera images. As the

lighting conditions in the real environment can

change, automatic adjustment of the threshold values

for the Canny Edge Detector is required. For this

reason, the two threshold values for the filtered Canny

algorithm are dynamically adapted. The median value

of the greyscale image is used for this normalisation

(Rosebrock, 2015).

Figure 4: Filtered Canny Edge Detector as used by the

model cars on the model motorway. Left: Perfect lighting

conditions. Right: Blurred vision because of car movement.

3.4 Distance Detection and Steering

Behaviour

For the model vehicle, the distance measurement to

the model vehicle in front is realised by an ultrasonic

sensor. The horizontal detection range of this

ultrasonic sensor is about 15°. This detection range is

sufficient for forming the emergency corridor, as the

vehicles autonomously keep a sufficient safety

distance and the rescue lane can therefore be formed

without extreme steering. Most importance has the

precision of the distance sensor. The ultrasonic sensor

in use sometimes detects false input pulses from the

surroundings, which lead to incorrect distance

measurement. This was also reported by ELEC

Freaks (2022).

During the development of the steering behaviour

of the vehicles, various tests revealed that the actual

pixel difference in the lane detection can be used to

control the steering. The pixel difference from the

centre of the lane to the centre of the model vehicle

can be used directly as the steering step in degrees.

IoTBDS 2024 - 9th International Conference on Internet of Things, Big Data and Security

324

Only a maximum steering threshold needs to be set.

We chose a maximum value of 10°.

3.5 Communication Between

Self-Driving Model Cars

Communication between the model vehicles is

established via wireless LAN (Wi-Fi). At this point,

it is the easiest way to realise communication, as the

model vehicles are already connected to Wi-Fi for

remote development. This simplifies the execution of

the various scripts on the Raspberry Pi and also makes

troubleshooting easier thanks to the Secure Shell

(SSH) network protocol. Also, the output of the

executed scripts can be monitored. Also, SSH access

to the model vehicle via Wi-Fi makes it easy to

calibrate the stepper motor and to configure the model

vehicles.

Communication between the model vehicles

allows them to react autonomously to various

messages. For example, as soon as an accident or

collision involving the model vehicle is recognised,

the vehicle performs emergency braking and notifies

the vehicles in the vicinity. The message contains a

name as well as information about the accident

vehicle's identification number. The various

messages of our specification are listed in Table 2.

Table 2: Overview of all possible messages including the

description for communication between autonomous model

cars.

Name Descri

tion

COLLISION Collision/rotation of the vehicle

FORWARD Forwardin

the incomin

messa

RESCUE

Notifying emergency vehicles of

an accident

LANE Own lane of the vehicle

TOTALLANES Total lanes of the motorwa

LOCALMAP Building the local map

POSITION

Own lane of the vehicle and ID of

the vehicle in front

STOP Stop own vehicle

CHANGESTATE Change of state

As already explained in Section 2, communication

between the model vehicles can be used to create the

required local map of the vehicles on the model

motorway to form an emergency corridor in the event

of a standing traffic. The LOCALMAP message is sent

to request the information from each individual

vehicle. The vehicles that receive this message

respond with the POSITION message. This message

contains information about the vehicle's own lane and

the identification number of the model vehicle in

front in the same lane. Each individual model vehicle

can then independently create its own local map with

all vehicles that are present in the real environment.

The CHANGESTATE message is sent to change the

state for executing the formation of the emergency

corridor (cf. Algorithm 1).

4 EXPERIMENTS

As before, we used our virtual simulator (Kuzmic and

Rudolph, 2020) to perform preliminary testing of our

new Algorithm 1. To proof that the results transfer to

the real world, we now attempt to carry out

experiments with our real model vehicles. This allows

the behaviour of the algorithm for forming the

emergency corridor in slow-moving and standing

traffic to be checked in reality. The different scenarios

are explained below using the illustrations. Videos

are also available online for all scenarios (Kuzmic,

2023).

4.1 Building an Emergency Corridor:

Slow-Moving Traffic

This experiment shows the algorithm for forming the

emergency corridor in slow-moving traffic with the

autonomous model vehicles, which are in the state

NORMAL_FOWARD. The emergency lane is formed

between the leftmost and the adjacent lane. When

forming the rescue lane, the position of the left or

right front wheel of the model vehicle must be

compared with the previously recognised lane

markings. At this point, the previously introduced

lane detection (Filtered Canny Edge Detector) from

Section 3.3 is used. This means that each model

vehicle knows in which direction (left or right) it has

to move in order to form a rescue corridor depend on

its own lane. As Figure 5, left, shows, the rescue lane

can be successfully formed by the autonomous model

vehicles.

Figure 5: Building an emergency corridor at slow-moving

traffic on a two-lane motorway. Left: Emergency corridor

about to open. Right: Emergency corridor opened and will

close if cars can speed up again.

In a slow-moving traffic, the model vehicles

automatically drive to the left or right to open a

Emergency Corridor Building on Multi-Lane Motorways with Autonomous Model Cars

325

corridor for emergency vehicles. If the speed of the

model vehicles increases, the opened emergency

corridor is automatically closed again (Fig. 5, right).

This ensures that the vehicles have formed a rescue

lane for the police and emergency vehicles at all times

after braking.

4.2 Building an Emergency Corridor:

Obstacles in Front

In this section, the formation of the emergency

corridor in the event of various obstacles on a

motorway is tested. If the speed decreases, it is

theoretically irrelevant for the model vehicles for

what reason it is. The position of an obstructive object

is more important than its type (e.g. crashed car,

crossing animal, etc.). As soon as an object is detected

in the vehicle's own lane and is approaching, the

speed of the model vehicle is reduced. Then, the

emergency corridor is opened automatically when the

speed falls below a certain value (e.g. 30 km/h). As

Figure 6 shows, the emergency corridor can be

successfully formed with the model vehicles on a

two-lane model motorway in the event of an accident.

Figure 6: Successfully established emergency corridor in

the event of an accident on a two-lane motorway. Left: Car

accident. Right: Animals crossing.

The rescue corridor can also be successfully formed

on a three-lane motorway in the event of an accident

with a closed lane. This is shown in Figure 7.

Figure 7: Successfully established emergency corridor in

the event of an accident on a three-lane motorway with a

closed lane. Left: Front-view. Right: Rear-view.

Additionally, the formation of the emergency corridor

with autonomous model vehicles can also be carried

out on a three-lane motorway. This is shown in Figure

8. In the previously published simulator, some

experiments were also successfully carried out on a

four-lane motorway with different lane widths

(Kuzmic and Rudolph, 2020).

Figure 8: Successfully established emergency corridor in

the event of an accident on a three-lane motorway. Left:

Front-view. Right: Rear-view.

4.3 Building an Emergency Corridor:

Front Tire Bursting

To test whether the emergency corridor can be

formed at any random time with the autonomous

model vehicles, the front tyre of the model vehicle

was simulated to burst. In this experiment, the impact

sensor was used to detect an unintended and sudden

movement of the model vehicle that we caused by

hand. The occurrence of this movement was

unpredictable for the model. Thus, the crashed

vehicle recognised its accident and performed an

emergency braking manoeuvre. As shown in Figure

9, left, this model vehicle is in the third lane (most

right lane in the picture). Once the accident and the

unforeseeable turn has been recognised, all vehicles

in the vicinity are notified in order to avoid collisions.

These following model vehicles can then

immediately react to the incoming message and

independently decide how to form the emergency

corridor.

Figure 9: Model realisation of a front tire bursting while

driving. Left: Front-view. Right: Rear-view.

The first model vehicle in the second lane also had to

perform an emergency braking, as this model vehicle

is close to the vehicle involved having the accident.

However, this first model vehicle in the second lane

was still able to form the emergency corridor already

as it steered safely during the emergency braking. All

IoTBDS 2024 - 9th International Conference on Internet of Things, Big Data and Security

326

other vehicles maintained the safety distance and

were therefore further away. This allowed these

model vehicles to brake normally and at the same

time open the rescue corridor for the police and rescue

vehicles (Fig. 9, right). The fact that such a complex

manoeuvre (simultaneous braking and targeted

steering) can be carried out is a decisive advantage of

autonomous vehicles, as it could never be performed

reliably by a human in such fractions of a second.

4.4 Building an Emergency Corridor:

Standing Traffic

This section presents the experimental tests of

Algorithm 1 for the formation of the emergency

corridor in a standing traffic in our real model

environment. For this, it is also necessary for each

individual model vehicle to generate a local map

through communication and optical evaluation of the

model vehicle in front. By exchanging its own

identification number, its own lane position and the

identification number of the vehicle in front, the

sequence of the model vehicles in the various lanes

can be created from the local map.

Figure 10: Building an emergency corridor in the event of

an accident in a standing traffic on a two-lane motorway.

Left: First row execution. Right: Third row execution.

This means that each model vehicle can make the

decision for itself which vehicles should start to form

the rescue corridor. The model vehicles

systematically form the rescue lane for the police and

rescue vehicles row by row. At this point, the

formation of the rescue lane in standing traffic takes

place via four states (see Algorithm 1). The model

vehicles closest to the accident vehicles start forming

the rescue lane (Fig. 10, left). If the distance is too

short for the emergency lane to be formed, the model

vehicles automatically drive backwards

(EXEC_BACKWARD). As our autonomous model

vehicles do not have rear sensors for distance

measurement, the vehicles behind notify the vehicles

in front of them of the distance. As soon as the model

vehicle in front has completed the formation of the

emergency corridor, it notifies the model vehicle

behind it in the same lane. From the third row

onwards, the distance for a rescue lane is sufficient so

that these model vehicles no longer have to drive

backwards (EXEC_FORWARD). This can be seen in

Figure 10, right. From the sixth row onwards, the

model vehicles are in the NORMAL_FORWARD state

and can build the emergency corridor in a slow-

moving traffic. As Figure 11 shows, the rescue lane

can be successfully formed on a two-lane highway in

a standing traffic.

Figure 11: Successfully established emergency corridor in

the event of an accident in a standing traffic on a two-lane

motorway. Left: Front-view. Right: Rear-view.

Also, the emergency corridor can be successfully

formed automatically on a three-lane highway. This

is shown in Figure 12.

Figure 12: Successfully established emergency corridor in

the event of an accident in a standing traffic on a three-lane

motorway. Left: Front-view. Right: Rear-view.

5 CONCLUSIONS

The algorithm developed for emergency lane

formation was successfully tested in reality using the

adapted lane detection (Filtered Canny Edge

Detector). The correct functionality of the algorithm

was demonstrated experimentally for slow-moving

and standing traffic in a real-world situation using

model cars. The model cars acted completely

autonomously. The reason and the external

circumstances for the formation of the emergency

Emergency Corridor Building on Multi-Lane Motorways with Autonomous Model Cars

327

corridor are irrelevant. This could be, for example,

slow-moving traffic, animal crossings, a burst front

tyre, a car collision or any other unforeseeable event.

Conducting the experiments with model vehicles

showed more difficulties than our previously

published simulation (Kuzmic and Rudolph, 2020).

In a real environment, the quality and measurement

tolerance of the sensors and the design of the model

vehicles for the test play an important role. The

previously published and in this paper extended and

improved lane detection (Filtered Canny Edge

Detector) could additionally be further optimised for

hardware with limited resources (Raspberry Pi 3 B).

However, about 30 FPS, which we achieved by CPU

parallelization, already enabled us to run our lane

detection in real-time.

In addition, selected examples of the experiments

in this work were published online (Kuzmic, 2023)

with references to our scientific publications.

For future work, the car body of the model

vehicles could be improved. It is also recommended

to use a better ultrasonic sensor for measuring the

distance to the vehicle in front. Finally, it would be

most interesting to test how the algorithm presented

in this paper performs in the real world with real

autonomous vehicles in real size on a real motorway.

REFERENCES

Austrian Federal Ministry of Mobility, 2023.

Rettungsgasse. Oesterreich.gv.at. [online]. Available

at: https://www.oesterreich.gv.at/themen/mobilitaet/kf

z/10/Seite.063130/Seite.065000.html. Accessed:

20.10.2023.

Buchenscheit, A., Schaub, F., Kargl, F., Weber, M., 2018.

A VANET-based emergency vehicle warning system,

2009 IEEE Vehicular Networking Conference (VNC),

pp. 1-8, ISBN 978-1-4244-5685-7.

Canny, J., 1986. A Computational Approach to Edge

Detection. In Proceedings of IEEE Transactions on

Pattern Analysis and Machine Intelligence, Bd. PAMI-

8. 6, pp. 679-698, DOI: 10.1109/TPAMI.1986.4767

851.

Colella, N., Herman, D. A., 2018. Emergency corridor

utilizing vehicle-to-vehicle communication. US-Pat.

US20170352268A1. Ford Global Technologies LLC.

ELEC Freaks, 2022. Ultrasonic Ranging Module HC -

SR04. Cdn.sparkfun.com. [online]. Available at:

https://cdn.sparkfun.com/datasheets/Sensors/Proximity

/HCSR04.pdf. Accessed: 17.02.2024.

Dębiński, M., Jukowski, M., Bohatkiewicz, J., 2018.

Emergency Corridors - Necessary Solutions in a

Modern Road Network, MATEC Web of Conferences

231 (GAMBIT 2018), pp. 1-6.

Kofler, M., Kühnast, C., Scherbeck, C., 2021. Raspberry

Pi: Das umfassende Handbuch. Galileo Press, pp. 17-

44, ISBN 978-3-8362-8353-3.

Kuzmic, J., 2023. Building an Emergency Corridor with

Autonomous Cars. YouTube.com. [online]. Available

at: https://www.youtube.com/watch?v=OEgJETouExc

&list=PLIzKtWM9Wfqen3TxLNnE-

noxu980jd_gy&index=14. Accessed: 17.02.2024.

Kuzmic, J., Rudolph, G., 2020. Unity 3D Simulator of

Autonomous Motorway Traffic Applied to Emergency

Corridor Building. In Proceedings of the 5th

International Conference on Internet of Things, Big

Data and Security (IoTBDS), pp. 197-204, ISBN 978-

989-758-426-8.

Kuzmic, J., Rudolph, G., 2021. Comparison between

Filtered Canny Edge Detector and Convolutional

Neural Network for Real Time Lane Detection in a

Unity 3D Simulator. In Proceedings of the 6th

International Conference on Internet of Things, Big

Data and Security (IoTBDS), pp. 148-155, ISBN 978-

989-758-504-3.

Marin, R. T., 1998. Emergency vehicle proximity warning

and communication system. US-Pat. US4794394A.

Stephens, R. S., 1990. A probabilistic approach to the

Hough Transform. In Proceedings of the British

Machine Vision Conference, British Machine Vision

Association, pp. 12.1-12.6, DOI:10.5244/c.4.12.

RaspberryPi, 2016. Raspberry Pi Camera Module 2.

Raspberrypi.com. [online]. Available at:

https://www.raspberrypi.com/products/camera-

module-v2/. Accessed: 15.12.2022.

Rosebrock, A., 2015. Zero-parameter, automatic Canny

edge detection with Python and OpenCV.

Pyimagesearch.com. [online]. Available at:

https://pyimagesearch.com/2015/04/06/zeroparameter-

automatic-canny-edge-detection-with-python-and-

opencv/. Accessed: 13.04.2023.

Siegel, M. A., 2003. Emergency vehicle alert system. US-

Pat. US20030043056A1.

SunFounder, 2022. Raspberry Pi Smart Car Kit (Picar-S)

for Intermediate. Sunfounder.com. [online]. Available

at: https://docs.sunfounder.com/projects/picar-s/en/sta

ble/introduction.html. Accessed: 28.11.2022.

IoTBDS 2024 - 9th International Conference on Internet of Things, Big Data and Security

328