Automatic Train Operation: History and Open Questions

Ale

s Lieskovsk

, Ivo Myslivec

, and Michal

Zemli

cka

1,2,3 a

avod Technika, A

ZD Praha, s.r.o.,

Zirovnick

a 3146/2, Praha 10, Czech Republic

Department of Informatics, Jan Evangelista Purkyn

e University,

Cesk

e ml

ade

ze 8,

Ust

ı nad Labem, Czech Republic

Department of Software and Computer Science Education, Charles University, Praha, Czech Republic

Keywords:

Automatic Train Operation, ATO Strategy, Human-machine Interface.

Abstract:

The paper presents the concept of automatic train operation. We give here short description of its functionality

and remember some points from its history. There is an overview of various future development as well as

proposals for improvement of some existing, especially mainline, solutions. There are presented also some

observations from decades of practical use of automatic train operation in the Czech Republic. Selected

challenges are presented and discussed.

1 INTRODUCTION

Even long-time used technologies as railway (what

could be expected to be already mature) could be fur-

ther improved. One of the possible improvements

what are slowly getting into action (or what is in-

tended to be used in relatively near future) is Auto-

matic Train Operation (ATO). It is a technology al-

lowing (under some circumstances) to ride a train. In

some cases (at special separated and protected tracks

as in underground or airport trains) it could replace

the driver, otherwise it could help the driver to fulﬁl

his hard work. It could help to care about speed lim-

its, stopping points, and to some extent also following

the timetable.

ATO itself is partitioned into two cooperating sys-

tems: one collects necessary data and distributes them

to the trains (usually called trackside), and the one

that could drive and control the trains (or take part on

it) using the received data (called onboard). The on-

board part usually cooperates with Automatic Train

Protection (ATP; a system that protects the train from

some critical situations like going too fast or running

through place where it is not allowed) and with train

control system.

The trackside part should collect and distribute all

the necessary data concerning train ride. They are col-

lected from various systems: track description from

asset management, timetables and current data from

(often multiple) Trafﬁc Management Systems (TMS)

that control trafﬁc within given area.

https://orcid.org/0000-0002-0951-0272

ATO can play multiple roles: It can simplify

driver’s work by taking control over speed, precise

stopping, and following the timetable. It can reduce

energy consumption by optimizing train speed. It can

reduce train and asset maintenance costs by reducing

the speedups and braking during the train ride. ATO

systems can also handle the door opening and closing.

For its proper use ATO must be able to recognize

where the train exactly is, what the limitations of the

track are, where are the planned stops, and a lot of

other information. There are two basic approaches to

equipping train with necessary information:

1. The train is always online and getting changes

whenever corresponding TMS collects them.

2. The train is equipped with track description and

timetable and potential changes could be handled

by informing the driver that could make some ad-

justments.

Let us make a tour on the prerequisites and history

of ATO, its current development and focus on its open

questions. We will mention several approaches and

compare them.

Structure of the paper is as follows: Overview of

principles used for ATO design and how they coop-

erate are in Sections 2 and 3. Data overview and re-

lated procedures are in Sections 4 and 5. Section 6

lists common challenges to be solved by any ATO

solutions. Section 7 compares on-line and off-line

ATO. Our experience from development and decades

of use of ATO are in Section 8. Common require-

ments of potential ATO systems are collected in Sec-

tion 9. Section 10 focuses on solution-speciﬁc chal-

260

Lieskovský, A., Myslivec, I. and Žemli

cka, M.

Automatic Train Operation: History and Open Questions.

DOI: 10.5220/0009418002600267

In Proceedings of the 6th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2020), pages 260-267

ISBN: 978-989-758-419-0

lenges. Our proposals and hints are collected in Sec-

tions 11 and 12. Finally, Section 13 summarizes the

conclusions.

2 BACKGROUND AND HISTORY

2.1 Technical Background

Some actions required for proper ATO functionality

are required also for proper functionality of ATP. As

safety is for railway more critical requirement than

simpliﬁcation of drivers’ work, some of the tech-

niques were developed primarily for ATP.

Such functions cover positioning of the train, en-

coding speed limits and signals, computing braking

curves, handling varying train length, and many oth-

ers. Many of the problems are complex and interest-

ing tasks also from IT perspective: For example, for

train positioning the input data are often taken from

various sources with varying quality but the result

(where to stop) must be quite precise – the train must

stop sooner than at indicated position (before ”stop”

signal or before end of track). Basic source of lin-

ear position (position on the track) is usually taken

from odometry (system computing current position

from wheel cycling). As the wheels could slip and

slide and their size is usually not known with abso-

lute precision, after a while such measurements could

differ from reality. It is therefore necessary to detect

the precise position using other means and restrict the

position error given by odometry. This precise posi-

tioning is performed by precise placement of balises

(special points detectable by trains) into track. Balises

usually work on magnetic or electromagnetic basis.

The balises must be (to some extent) unique – they

must be equipped by a code identifying them. It al-

lows the trains to recognize where they are.

Global satellite positioning systems (like GPS,

Galileo or GLONASS) are currently not always pre-

cise enough to deliver the data according to the needs

of ATO. For use in railways the shift of three meters

may cause erroneous positioning of the train to paral-

lel tracks. But even parallel tracks could have differ-

ent speed limits and different positions of signals and

platforms.

Satellite positioning is acceptable for lengthwise

positioning (where on the given track the train is) es-

pecially at local tracks. Precise lateral positioning

must be supported by additional systems or by human

involvement.

If the train should stop with given precision, the

track data must be available with at least slightly bet-

ter precision.

The railway speed limits are related not only to the

train front but also to the train rear – usually it applies

what is more restrictive. It is, the train (its driver as

well as its ATP and ATO systems) ”must know” its

length to identify, what are the real speed limits along

the track.

The above mentioned techniques and procedures

developed for automatic train protection could be of-

ten used (after some changes) for the automatic train

operation. Some of them have been directly devel-

oped for ATO: when an older ATP system, which does

not provide localisation information, is used, ATO re-

quires its own source of precise train positioning.

2.2 Highlights from ATO History

First systems helping drivers to stop by the platform

were developed in the second half of XX-th century.

The ﬁrst ATO systems were developed for closed

railway systems like metro or special railways. The

Victoria Line in London (UK) has been opened with

automatic train operation in 1968. The Bay Area

Rapid Transit using multiple lines ATO has been

opened in 1972. Line C of Prague Metro has been

equipped with ATO in 1978.

The ﬁrst ATO used for classic railway in everyday

operation (AVV, automatick

e veden

ı vlaku) has been

introduced in Czech Republic in 1991. It works on

many lines (around 3000 km of total approx. 9000 km

od Czech Republic railway network length) running

on shared tracks with non-ATO trains.

2.3 Current Development

During last years there is a signiﬁcant development in

ATO driven by various projects in various countries.

Some systems are based on tight cooperation with sin-

gle ATP (e.g. ATO over ETCS), others are going to be

more open.

There appear ﬁrst demonstrations or even applica-

tions of fully automated (unattended – without driver)

classic railway ATO in many countries (Australia,

China, Russia, ...). For practical use it makes sense to

make them interoperable in larger areas (huge coun-

tries or even continents).

Later in this paper we will mention some of the

prerequisites of such interoperability.

3 HOW ATO WORKS

As written above, ATO is divided into two basic parts

(compare also Fig. 1):

Automatic Train Operation: History and Open Questions

261

TMS – Trafﬁc Management System



Track map

database

ATO trackside

ATO on-board

ATO

Figure 1: Rough view on ATO and cooperating systems.

1. Trackside – collecting and converting data from

various sources to the form usable for the second

part

2. Onboard – performs following actions:

(a) positioning the train;

(b) computing dynamic speed proﬁle according

track description, signals, additional restric-

tions, train attributes, current conditions, and

timetable;

Trackside part of ATO must be able to cooperate with

other trackside systems – especially with TMS that

have timetables and asset management that keeping

track description up-to-date.

Building trackside part is a task of infrastructure

manager, whereas equipping trains with the on-board

part must be done (ordered and paid) by the vehicle

owner. As many trains are going across borders be-

tween areas under control of different infrastructure

managers and within areas of individual infrastructure

managers There are several projects trying to make

the necessary parts interoperable across several coun-

tries – equipping tracks and vehicles by ETCS. The

proposed standardized architecture is – with respect

to work of ATO WP in Shift2Rail published e.g. in

(Buurmans, 2019) or in (X2Rail-1, 2017) – in a sim-

pliﬁed form captured in Fig. 2. Subsets are speciﬁca-

tion documents of the ERTMS/ETCS system used in

the European Technical Speciﬁcations for Interoper-

ability of the Railway System.

4 NECESSARY DATA

Complexity of the task can be derived from the data

that are necessary. The following simpliﬁed list con-

tains data for both (on-line and off-line) ATO ver-

sions:

• Track description

– speed limits according train class (position,

value, and front/rear validity);

– gradients (value, position);

ground systems

TMS – Trafﬁc Mgmt System

subset

131

track data

ATO trackside

-

subset

126

ETCS on-board

subset

034 or 119

subset

-

130

ATO on-board

subset

139

basic train control

















Figure 2: Cooperation of ATO with related services and

their speciﬁcation in subsets.

– platforms (positions, length, train alignment);

– position of signals (if not transmitted from

ATP);

• timetable

– times for planned stops

– times for passing points (time navigation)

• other limitations

– power limits

– axle load limits

– ...

5 HANDLING DATA CHANGES

The data could change even when the train is on the

track – additional limits could appear, the train will

take other path, etc. The issue is, how the train can

cope with it.

The data update could be sent to the train only if

it is available online. When the train is not connected

to the network, other approach must be used.

5.1 Connected Train

European Union Agency for Railways (ERA

) plans

to introduce a standardized communication between

TMS and ATO and between ATO trackside and on-

board parts based on the work performed by the com-

panies organised in Shift2Rail

, UNISIG

and EUG

https://www.era.europa.eu/

https://shift2rail.org/

http://www.ertms.net/?page id=50

https://ertms.be/

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

262

5.2 Off-line Train

The selection of different path (using other track than

planned but still going through the same stations)

must be always taken into account: the train speed

must reﬂect the current strongest speed restriction. In

some cases the selected routes could be recognized

from signalling: straight track usually does not re-

strict the speed (and could be signalized by ”free”),

whereas siding often limits the speed (and could be

therefore signalized by lower speed limit). The in-

formation about the active signal could be transferred

through ATP.

6 COMMON CHALLENGES

The way the data are reaching the train does not in-

ﬂuence some of the technical challenges that must be

solved anyway. It covers train positioning, speed lim-

its, braking curves, acceleration, and many others.

Many of the challenges have been already dis-

cussed in literature (and, of course, tested in practice).

Let us mention at least some of them.

6.1 Train Positioning

One of the hardest challenges for ATO is precise, safe,

and reliable train positioning. Although trains are

moving on rails (and their positions are known) it is

still not easy to position the train precisely enough:

GNSS (global navigation satellite systems) localisa-

tion error may be greater than the Also the positioning

on the track is often required to be more precise than

the GNSS-based position could give.

One can say that automatic train protection (ATP)

systems must have precise enough positioning. They

can determine the track. But they have different ap-

proach to linear positioning: For an ATP system the

error could be asymmetric – if it stops several (proba-

bly also several dozens) meters sooner than necessary,

it is still good enough. But it is not allowed to stop

even a bit later.

ATO has other requirements on train positioning:

it needs to stop not too late but also not too soon –

e.g. not to miss the platform or not to speed up too

soon.

6.2 Speed Limits

In some cases the speed limits concern front of the

train, in other cases its rear. In most cases it holds

for the more restrictive case: when the speed limit

is getting more restrictive, it must be fulﬁlled by the

train front (front of leading vehicle), whereas speed-

ing up must be usually done with respect to the rear of

the train. Exceptions exist and are marked by special

signs allowing immediate speed up.

Some speed limits are ﬁxed for a long time (typ-

ically the ones related to the physical parameters of

the track). These parameters could be distributed to

the trains on-line as well as off-line.

Temporary speed restrictions and speed restric-

tions given by route setting are of dynamic nature and

need to be updated immediately. They must be trans-

ferred on-line or entered by operator (driver). For (au-

tomatic) driving it is only important whether the data

are available in time. From the ATO perspective it is

not important how the data get into the system.

As mentioned above, it is important that the train

must reliably know its length. Although it could look

easy, it is not so. Trains could be composed not only

from new vehicles but also from older ones – even

decades old. Such vehicles need not be necessarily

equipped by the most current electronics and soft-

ware.

6.3 Acceleration and Braking Curve

If we know (as described above) the speed limits, we

should also drive so that we obey them. There are

many parameters that should be taken into account:

• gradients

• tunnels and their diameter/tightness

• adhesion

• available power output and its limitation

• braking type and mode

• ...

The above parameters could inﬂuence the train indi-

vidually as well as in combination. Some of the pa-

rameters are not known precisely – so, the control

must therefore take into account also the current be-

haviour of the train.

7 DIFFERENCES

It appears that on-line and off-line ATO systems can

share most algorithms – computations like position-

ing, speed proﬁles, timetable matching, braking, or

door opening appear to be the same. The differences

are in availability of information about surrounding

trafﬁc, and real route settings. On-line ATO can have

all such information – if the connection with its track-

side part will work as expected and everywhere where

needed.

Automatic Train Operation: History and Open Questions

263

The main difference comes when the systems

come under stress: On-line systems could take ad-

vantages from updates if they are prepared and sent in

time by trackside (or the TMS behind). Off-line sys-

tems have only pre-loaded data. It is therefore mean-

ingful to load them with full maps and equip the track

so that they could quickly recognize where the route

leads (to which speed proﬁle and potential stops).

On-line ATO can result in more efﬁcient ride of

the trains (as they have more up-to-date and more pre-

cise data) – but only where the data are available and

in time. Off-line ATO is cheaper and can be used even

when the surrounding trafﬁc data is not available.

8 EXPERIENCE FROM AVV

8.1 A few Gossips from AVV History

Automated Train Operation/Control (In Czech: Auto-

matick

e veden

ı vlaku (AVV); (Myslivec et al., 1998;

Lieskovsky and Myslivec, 2011)) is a system in-

stalled on hundreds of vehicles and used in practice

for many years (in commercial operation since 1991,

tests since half of 1960’s). Its basics have been de-

veloped decades ago. It has been therefore limited to

preloaded data available on board (at those times no

GSM-R or related technology has been available).

The dependency on preloaded on-board data led to

the graph-like track description. The graph contains

all tracks equipped with AVV-compatible balises. The

system can be therefore used everywhere where there

are necessary data available and where it is possible

to position precisely the train. The system is there-

fore robust to any change of the train path within this

graph. It the train is sent to some part of the track

where no information is available, it switches to con-

trol by the driver. The system always takes into ac-

count all (known) possible paths.

8.2 Cooperation of AVV with ATP

The AVV system reads the signals from national Class

B ATP system (compare (Lieskovsk

y et al., 2006))

and is capable to utilise data from onboard ETCS

if available (see (Lieskovsky and Myslivec, 2011)).

Even if no ATP is available, AVV can work alone and

can prevent some critical situations (can slow down

or stop the train if necessary).

8.3 AVV Approach to Data

The system is able to work without connectivity to

TMS On the other hand, it is limited to the areas for

whose it is equipped with corresponding maps. The

system is designed so that if a driver has some useful

information (s)he can enter the data relatively easily.

9 COMMON REQUIREMENTS

It is likely, and the available documents prove, that all

ATO systems need some basic information. It covers

information about speed limit start and end positions,

their validity for beginning or end of the train; plat-

forms and their positions; stopping points (their po-

sitions and required train alignment there); and train

timetable. There should be also information identify-

ing stations (to provide information about them to the

passengers), optionality of the stop (to induce its po-

tential request). Additionally, there should be avail-

able also some safety requirements (not to stop on

some places where it is not possible to rescue people

from the train).

If possible, it can be advantageous if there is in-

formation about current conditions around the train

like signals or information about other trains (e.g. if

there is a slow train ahead of given train, it could be

reasonable to slow down a bit); sometimes it could be

advantageous to speed up to allow smooth train cross-

ing on a more appropriate place.

The data could be stored efﬁciently. The detailed

track description covering haul between Praha and

Kol

ın (62 km distance, multiple stations, about 320

km of single track, single direction line) have been

stored in a 3KB piece of EPROM.

10 DIVERGING REQUIREMENTS

Various ATO systems could require different addi-

tional information. It can cover position and radius

of curves, some could require also their direction (in

other cases only modiﬁed train resistance could be

enough). Similarly, some systems can take advantage

from information about tunnels and bridges and their

parameters.

As it seems to be advantageous to have this in-

formation as in these places, the train running resis-

tance really is higher than the one in straight track in

open country, the fact is that the running resistance

provided by vehicle manufacturer is usually higher

than reality (not to be a problem to verify it during

the tests) and after adding this additional component,

the expected running resistance goes more away from

reality... Finally, there exists a component that is not

possible to take into computations: wind. Its asset

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

264

to running resistance is much higher than the one of

tunnels, bridges or curves.

In some countries it is allowed to stop at a long

enough platform only, in some countries there is

sometimes allowed to stop at a short platform and

open appropriate doors only. Support for it could

be also recognized as useful in some cases only

(e.g. some infrastructure managers could require it

and some do not want to care about).

It could be useful to distinguish how the doors are

opened: at some stations they could be opened au-

tomatically, on other stations they could be only un-

locked for opening on demand.

11 THE PROPOSAL

11.1 Approach to ATO

We propose to treat ATO as a general service with

preference of robustness over implementation sim-

plicity.

We suppose that customers could beneﬁt from

having ATO as able to work on ETCS- as well as

national ATP-equipped infrastructure. Having infor-

mation about the planned path with adjacent tracks

(i.e. to have the information of the track not only as

a line but as a general graph) could make the system

more robust to the data failures.

ATO solutions capable to operate on both interop-

erable (ETCS) and non- interoperable lines can ben-

eﬁt from the data from various ATPs. It can ex-

tend applicability of the ATO technology with all its

advantages to signiﬁcantly more tracks and trains.

Measurements published in (

Sula, 1990) and men-

tioned again in (Jackson, 2017) has shown that AVV

has saved 10–30% of energy as well as improving

timetable. Can we say that these advantages will not

be allowed on local tracks?

11.2 On-line vs. Off-line

We suppose that the modern ATO systems should be

built so that they can work in both environments: on-

line as well as off-line. It makes sense to build the

ATO system so that it could be used on-line as well

as off-line and with any ATP fulﬁlling some reason-

able minimal requirements. It also makes sense to

deﬁne minimal set of functions necessary for driving

the train by ATO. Such interface could then be used by

ATO without necessity to re-certiﬁcate the train con-

trol in all countries.

11.3 Opportunity not to Collect Unused

Data

Some data are in some countries already not col-

lected. As they are typically not used there, it could

be enough to set up generally some default value and

do not mention it in all track-describing datagrams.

But it requires the applicability of country- or area-

predeﬁned values and more variable data format.

On the other hand, if some data are not available,

it could be better to have an opportunity to support

only partial description of the track. If the on-board

ATO can use the available data, OK, if not, let it (tem-

porarily) switch to manual control over the train.

11.4 ATO–train Interface

The communication between ATO and vehicle there

must be at least:

• request on traction / brake;

• backward information on the successful use of

traction / brake;

• other parameters of the train not processed by

ATP (often ETCS) – like weight, overall power,

or traction limits.

11.5 ATO–ATP Interface

The data transfer between ATP and ATO contains the

data that is ATP able to provide. In the case of ETCS

it is especially:

• static speed proﬁle;

• braking and other parameters of the train

(e.g. train length) to synchronize the values be-

tween ETCS and ATO; ATO is therefore able to

compute the same braking curve as ETCS (and

anticipate the braking activities of ETCS);

• localization data (identiﬁcation of the position of

the train on the track / network);

• last but not least, data from ATO for displaying

on the ETCS display. It appears to be reason-

able (and about 20 years of practice with AVV

shows that it is gracefully accepted by the drivers)

to provide extra display with complex information

about the situation on the track in front of the train

and about plans of the ATO itself.

11.6 Data for ATO

The data for ATO system can be divided into follow-

ing groups:

Automatic Train Operation: History and Open Questions

265

1. Static (changes usually after several years or

decades) – track description, speed limits, po-

sitions of platforms, points of interest, stopping

alignments, etc.

2. Semi-static (could change several times a year) –

published timetable (train routes, stops, departure

and arrival times, ...)

3. Varying (usually change in weeks or months) –

temporary train path changes (e.g. due reconstruc-

tion of the track), temporary speed restrictions, ...

4. Dynamic (could change during the train run) –

selected routes, time restrictions (the train should

be there not sooner than t

and not later than t

12 SEMI-OPEN QUESTIONS

The world of ATO faces many questions. Some of

them have been handled in the past but current solvers

must try it again. Let us mention and formulate some

of them and give some hint according the ﬁrst try.

12.1 Optimal Human–Machine

Cooperation

For the ATO community it is not clear what the op-

timal way of human-machine (driver–ATO) coopera-

tion is. Some authors (compare e.g. (Bienfait et al.,

2012)) claim that there must be clear cut between

full control by machine (ATO) and full control by the

driver. Others (compare e.g. (Lieskovsky and Mys-

livec, 2011)) propose to combine advantages of both

worlds. Practical combination is probably more com-

plicated to implement (and, what is harder, requires

deep understanding of the designers/developers of the

interface to train driving) but appears to be less re-

strictive and more ﬂexible.

Let us formulate it using following parable: The

driver is smart and experienced enough to optimize

train driving based on his past experience and has

a better access and understanding of speciﬁc opera-

tional rules and limitations. The machine could be

better than the driver as it could have more detailed

or more precise speciﬁc data regarding train drive and

due to fast computation algorithms can provide spe-

ciﬁc action more precise and faster. Symbiotic work

of the machine driver and computer can take advan-

tage from both worlds: from precise computation over

lots of date and from experience and understanding.

12.2 Data Inconsistency Handling

As ATO as a whole collects for its proper work in-

formation from many systems and as (according cur-

rent knowledge from physics) the speed of a signal

is limited, it could happen that the data from vari-

ous sources are not synchronized. The question is,

whether we should expect that the data synchroniza-

tion is the task of all connected systems (especially

TMS) as it is expected in current developments in

ATO standardization in Europe or should there be

mechanisms in data distribution and exchange allow-

ing to recognize possible inconsistencies.

It seems that it is safer to be prepared for situa-

tions where something went wrong. We propose to

sacriﬁce small part of the system data throughput for

mechanisms allowing recognizing of potential issues

just to allow reasonable reaction.

12.3 Communication to Driver

What is the optimal extent of the information that

ATO should share with the driver? Should it be mini-

mized (not to disturb the driver) and safe a display or

should there be complex information on the situation

ahead and on the plans of ATO? The currently pro-

posed solution in AoE (one line of text is enough) is

derived from the fact that the ATO should not cooper-

ate with driver but drive the train alone (until the train

stops or until the driver takes control over the train).

The approach used in AVV (compare Figs. 3 and 4)

prefers providing complex information the driver by

ATO with all necessary data (like how far is the next

speed limit, what is the next signal and where it is, or

what is the next stop).

This picture was created upon of opinions and

notes of approximately 100 drivers, which took part

in the ﬁrst year of real computer-based ATO opera-

tion on Czech Railways in 1991. In contrary to other

approaches, it comes from real, not simulated opera-

tion, and from real drivers, who were responsible for

safety and punctuality of real trains. After implement-

ing changes, this picture remains stable for almost 30

years of everyday use.

13 CONCLUSIONS

We have described automatic train operation as useful

intelligent system facing many interesting challenges

and issues. It has already shown its practical advan-

tages – compare its use in underground in many cities,

in several airport local trains, as well as in regular use

in standard railway (AVV).

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

266

Figure 3: AVV Display Explained.

Figure 4: Diagnostic and AVV Displays.

Several challenges have been introduced and dis-

cussed as well as current developments in the area. It

has been observed that a signiﬁcant group of current

developers tries to re-solve identiﬁed challenges that

have been already put and answered in the past. We

have described them and have tried to explain why the

questions have been answered so.

Currently the practical task in the ATO is to equip

the railway with the necessary equipment (balises),

to equip the trains with corresponding equipment, to

collect the necessary data and to connect the ATO to

other trackside systems. It appears that many ques-

tions of ATO concept and implementation (some of

them were mentioned here) could be answered only

if the corresponding people will get their own experi-

ence with ATO development, testing, and use.

REFERENCES

Bienfait, B., Zoetardt, P., and Barnard, B. (2012). Auto-

matic train operation: the mandatory improvement for

ETCS applications. Paper presented at Aspect 2012.

Buurmans, K. (2019). Automatic train operation over

legacy automatic train protection systems : A case

study on the Groningen-Buitenpost line. Master’s the-

sis, Delft University of Technology, faculty of Civil

Engineering and Geosciences, department Transport

& Planning.

Jackson, C. (2017). In search of the game changers. Rail-

way Gazette International, 173:29–32.

Lieskovsky, A. and Myslivec, I. (2011). ATO and ETCS

can work together. Railway Gazette International,

167(3):37–40.

Lieskovsk

y, A., Myslivec, I., and

Spa

cek, P. (2006).

ETCS a AVV – spolupr

ace, nikoliv konkurence.

edeckotechnick

y sborn

ık

CD, (21):1–6.

Myslivec, I.,

Spa

cek, P., and

Sula, B. (1998). Automat-

ick

e veden

ı vlaku AVV. V

edeckotechnick

y sborn

ık

CD, (5):29–38.

Sula, B. (1990). Einﬂuss der Fahrtechnik auf den Energie-

verbrauch eines Personenzuges. Elektrische Bahnen,

(4).

X2Rail-1 (2017). ATO over ETCS (up to GoA4). Presented

at CRCC 2017.

Automatic Train Operation: History and Open Questions

267