Automated

container terminals
and self-driving cars:
Industry
outlook
AUTHORS

JARI HÄMÄLÄINEN
PEKKA YLI-PAUNU
PERTTI PEUSSA
 Contents
AUTOMATED CONTAINER TERMINALS

1. EXECUTIVE SUMMARY 3

2. DRIVERS FOR AUTOMATION AND EXPECTED IMPACT 4

AND SELF-DRIVING CARS

3. TERMINOLOGY AND DEFINITIONS 6

4. WHERE WE ARE NOW 12

5. TECHNOLOGY CONSIDERATIONS 15

6. SAFETY 24

7. FUTURE PROSPECTS 28

8. SUMMARY 34

2
Executive summary
In recent years, the development of autonomous cars has progressed
rapidly, while automation has increasingly been gaining ground
at selected container terminals. This white paper outlines the key
differences and similarities between the two fields and examines
the influence that the development of autonomous road vehicles is
expected to have on the future of container terminal automation.

In both autonomous cars and terminal automation, the level of

automation is increasing stepwise. The two areas share many common
drivers towards automation but differ markedly in several respects.
The underlying operating differences in basic technical approach (most
notably that of autonomous road vehicles vs. a centralised terminal
infrastructure) will remain for the foreseeable future.

Arguably the single most important long-term enablers for autonomous

road vehicles will be Artificial Intelligence (AI) and especially machine
learning, which are the core technologies required for accurate real-time
situation analysis and safe decision making in complex open traffic
environments. It is the view of the authors that the exponential growth
of these technologies in the car and Information & Communication
Technology (ICT) industries will create significant openings for adapting
and integrating new capabilities into terminal automation, thus speeding
up the development of our industry.
 1. Drivers for automation and
expected impact
AUTOMATED CONTAINER TERMINALS

SCOPE OF DISCUSSION
AND SELF-DRIVING CARS

This white paper is intended as a high-level overview on the current

state of autonomous vehicle development and terminal automation – as
well as the differences and potential synergies between the two – for
readers in the container handling and logistics industries. Automation
is progressing step by step in container terminals as well as in the car
industry. However, in both fields, this is not just incremental product
development, but a major transition that will likely have a substantial
effect on the world at large.

At the time of writing (late 2018), both areas are undergoing fast
development, with autonomous vehicle research benefitting particularly
from recent progress in powerful and accessible AI / machine learning
capabilities. Instead of focusing on individual solutions, products
or research projects by various manufacturers, the paper aims to
facilitate analysis and discussion of how advances in self-driving vehicle
development might benefit the container handling industry, as proven,
cost-effective technical solutions become available for customisation
and adaptation.

THE CAR INDUSTRY AND SOCIETY

Over the last several years, manufacturers have been introducing

progressively more advanced driver assistance features into mass-
production cars. The purpose of these technical features is to aid the
driver in their journey from A to B while making driving more enjoyable
and safe. In this way, car manufacturers develop their products
incrementally in order to keep the industry and their business evolving.
The next goal is to increase the productivity of the driver or society. At
Automation levels of automation where no driver is needed, people will be able to
spend their time on something more productive or interesting during the
is likely journey. Once the majority of cars are self-driving, this can be expected
to have a to have a major societal impact.

substantial An even more significant effect on society and global business may
effect on result from various kinds of mobility as a service (MaaS) offerings. For
example, driver salaries are often the single biggest expense for taxi and
the world at truck companies. Once equipped with self-driving fleets, their business
large. is expected to get a major economic boost.

4
The biggest Also, self-driving cars will likely become a reality in MaaS applications
much sooner than they are generally available to consumers all around
societal impact the world. For example, self-driving public transportation, valet parking

of autonomous and local taxi services can be offered in selected areas when the
weather allows long before global fleets of self-driving cars are feasible.
vehicles may Development in this field is already well underway; in 2018, former

ultimately Google company Waymo announced the order of 82,000 vehicles

from Chrysler and Jaguar for a large-scale pilot programme of an
result from a autonomous commuting service in Phoenix, Arizona.

new type of Paralleling the development of MaaS offerings, the biggest societal
sharing impact of autonomous vehicles may ultimately result from a new type of

economy. sharing economy. Once the consumer does not have to own their car,
but can get a drive when needed (from a fleet of self-driving cars), the
number of cars worldwide can be expected to decrease significantly.
This may ultimately impact a diverse range of seemingly unconnected
businesses and services. Even the earnings model of car manufacturers
may change from selling vehicles to a service-based model in which
customers pay for kilometres travelled.

CONTAINER TERMINALS AND THE GLOBAL LOGISTICS CHAIN

Currently, automation is accepted as the primary way for terminals to

develop their competitiveness in the future. The benefits of automation
include safety, reliability, predictability and improved operations. Terminal
automation as such does not radically change the current business
models of the logistics industry, but the ever-growing software and
service business, as well as the development of open and transparent
global logistics chains does do so.

The increase of automation levels at container terminals parallels the

Terminal automatisation that has taken place in many other industries over the

automation last decades. In addition to the obvious safety improvements that arise
from keeping people out of the operating area of heavy machinery,
sets the stage automation enables terminals to do more container moves with the

for a fully available number of people. When combined with the rapid advance
of digitalisation and new developments such as blockchain, terminal
transparent automation sets the stage for a fully transparent global logistics chain

global logistics that can be expected to have a major disruptive impact on how goods
and materials are transported and delivered worldwide, and how
chain. businesses make a profit from it.
 2. Terminology and definitions
AUTOMATED CONTAINER TERMINALS

KEY TERMS

The Society In a discussion of automated vehicles and equipment, it is useful to

provide some clarification of the basic concepts in use. In everyday
of Automotive
AND SELF-DRIVING CARS

language, terms such as automated, automatic, autonomous, highly

Engineers automated and self-driving are used interchangeably; however, more
careful definitions are usually required when addressing the specific
(SAE) defines capabilities and development possibilities of such systems.
six progressive
The automotive industry usually uses the terms autonomous or self-
levels of driver driving when referring to the ultimate design goal of creating road
assistance vehicles that would eventually be able to handle any driving situation as
well as or better than a human driver. The United States Department of
technology. Transport refers to the same concept with the term highly automated
vehicle (HAV).

By contrast, the container handling industry deals with terminal

automation and automated container handling equipment, with little
foreseeable need to create fully autonomous vehicles in the future.
Furthermore, it is important to note that various levels of automation can
be deployed to support operational and business goals, but they do not
necessarily need to lead to fully autonomous vehicles.

To gain a more nuanced understanding of current solution levels and

future development prospects, it is worth examining the definitions of
progressive automation levels for both of these areas.

LEVELS OF DRIVING AUTOMATION

The Society of Automotive Engineers (SAE) defines six progressive

levels of driver assistance technology, from manual driving to fully
autonomous operation:

Level 0
The human driver does all the driving.
Level 1
An advanced driver assistance system (ADAS) on the vehicle can
sometimes assist the human driver with either steering or braking/
accelerating, but not both simultaneously.
Level 2
An advanced driver assistance system (ADAS) on the vehicle can
control both steering and braking/accelerating simultaneously under
some circumstances. The human driver must continue to pay full
attention (“monitor the driving environment”) at all times and perform the
rest of the driving task.

6
THE 5 LEVELS OF DRIVING AUTOMATION

HUMAN DRIVER AUTOMATED SYSTEM

Steering and Monitoring Fallback when Automated

acceleration/ of driving automation system is in
deceleration environment fails control

NO
1 AUTOMATION
N/A
monitors the road
Human driver

SOME
DRIVER
2 ASSISTANCE
DRIVING
MODES

SOME
3 PARTIAL
AUTOMATION
DRIVING
MODES

SOME
4 CONDITIONAL
AUTOMATION
DRIVING
MODES
Automated driving system
monitors the road

SOME
HIGH
5 AUTOMATION
DRIVING
MODES

FULL
6 AUTOMATION

Source: Society of Automotive Engineers / National Highway Traffic Safety Administration

https://www.nhtsa.gov/technology-innovation/automated-vehicles-safety
 Level 3
An Automated Driving System (ADS) on the vehicle can itself perform
AUTOMATED CONTAINER TERMINALS

all aspects of the driving task under some circumstances. In those

circumstances, the human driver must be ready to take back control at
any time when the ADS requests the human driver to do so. In all other
circumstances, the human driver performs the driving task.
AND SELF-DRIVING CARS

Level 4
An Automated Driving System (ADS) on the vehicle can itself perform all
driving tasks and monitor the driving environment – essentially, do all the
driving – in certain circumstances. The human need not pay attention in
those circumstances.

Level 5
An Automated Driving System (ADS) on the vehicle can do all the driving
in all circumstances. The human occupants are just passengers and
need never be involved in driving.

8
The AUTOMATION LEVELS IN CONTAINER TERMINALS

automation In container terminals, automation is also progressing stepwise from

level can be traditional manual operation through various process automation,

remote control and operator assistance solutions, to eventual fully
selected automated operation. The desired automation level can be selected

based on based on the existing systems, operating environment and business

goals of the terminal. Typically, process automation is deployed prior to
the existing – and alongside – automated equipment as the first step in gaining the

systems, benefits of terminal automation.

operating Kalmar defines several progressive levels of automation for container

environment handling equipment. It should be noted that the levels detailed below
are indicative and based on an application of an automated rubber-
and business tyred gantry crane (AutoRTG) system. However, they are generally

goals of the applicable to various other types of container handling equipment.

terminal. Level 0: Manual operation with process automation

In the traditional operating mode, a human driver controls all equipment
functions. Process automation adds another dimension to terminal
automation, as it can be deployed prior to, or alongside, different
automation levels for the actual equipment. Even with a human driver
still controlling the equipment, terminal performance can be improved
with various process automation, fleet management and operator
assistance solutions.

Examples of process automation solutions include automatic

identification and tracking of containers and trucks in the yard;
automated job allocation to equipment operators; telemetry for
horizontal transportation equipment; and stack profiling for preventing
collisions between the crane spreader and the container stack.

Level 1: Remote control

The most basic level of terminal equipment automation is remote
control, which already improves efficiency and safety by bringing
operators from the container yard into a control centre. For many types
of equipment (e.g. RTGs) remote control enables a single operator
to control multiple cranes. Remote control provides the possibility of
optimising the manning level of the terminal based on the true amount
of moves needed, instead of the number of operational cranes, as the
operators are located in an office environment at remote control desks,
able to take control of any crane in the terminal. Remote control also
offers other benefits such as eliminating the time needed for operators
to commute to and from the cranes in the yard.
 Level 2: Supervised automatic moves
At the next level of automation, crane operation is fully integrated
AUTOMATED CONTAINER TERMINALS

with the Terminal Operating System (TOS), so jobs are handled in an

integrated solution and not in a separate equipment-specific system.
Connection of the control desk to the crane is automatic once the job
arrives. At this level of automation, an RTG crane executes automatic
AND SELF-DRIVING CARS

gantry and trolley moves to the target location, as well as automatic

hoisting in the stacking area, all under the operator’s supervision.
In many cases, supervised operation makes operation faster while
reducing the potential for collisions against other containers. As the
crane automatically positions itself accurately over the target container,
this saves valuable seconds on each move. The automated moves
are also smoother, resulting in less wear and tear on equipment
and containers. This level of automation (supervised operation
with automatic gantry steering) is the desired option for new RTG
installations.

Level 3: Semi-automated operation

More advanced automation features can be added for higher efficiency
and performance, and at this level, remote operators can control even
more cranes per person. In semi-automated RTG operations, the
crane executes automated trolley movements in addition to automated
container pick and placing in the stack area. This level of automation
significantly decreases the required operator time per container move.
Gantry movement takes place under operator supervision, while
In a fully truck lane operations and exception handling are accomplished by

automated remote control. Trolley, hoist and gantry movements as well as stack
housekeeping can also be automated, with the operator remote
terminal, all controlling the crane only in the truck lane.

crane Level 4: Fully automated operation

functions Finally, in a fully automated setup, all crane functions including hoist
operation, container picking and placing, gantry moves and truck
including hoist lane operation are automated. If needed, an operator can still step
operation, in to manage exceptions remotely. With a fully automated system,
operator time per crane move is minimised, and stack housekeeping
container can be fully automatic. This level is standard for automated container
picking and handling equipment that can operate in a fully segregated automation
area without people. Examples include automated stacking cranes
placing, gantry and Kalmar AutoStrad™ straddle carriers that have already been in
moves and operation at various terminals for years.

truck lane
operation are
automated.

10
Automation levels in an AutoRTG terminal
Note: The depicted automation levels are illustrative and specific to an automated rubber-
tyred gantry crane installation. Exact functionality at increasing levels of automation will
depend on the type of automated container handling equipment and terminal concept.
 3. Where we are now
AUTOMATED CONTAINER TERMINALS

AUTOMATED CONTAINER TERMINALS TODAY

Automated container terminals have already been in production use for

AND SELF-DRIVING CARS

over two decades. The world’s first automated terminal was ECT Delta
in the Netherlands, operational since the early 1990s. It was followed
by the HHLA CTA Terminal in Hamburg, Germany in 2000. In 2005, the
first stage of Patrick’s AutoStrad™ straddle carrier terminal went live at
the Port of Brisbane, Australia. Over the years, Patrick has continued
to provide high performance levels while simultaneously becoming
one of the world's safest container terminals. Today, some degree of
automation is the standard for all newly built container terminals.

There are several reasons why container terminal automation has, until
now, developed further than roadgoing autonomous vehicles. Firstly,
the operational area is limited to the terminal domain, which simplifies
deployment. Secondly, a terminal site is always relatively flat and it is
easy to manage the full infrastructure within the area, from fences,
gates and lamp masts to traffic signs, communication, cabling, and
navigation markers. The majority of container terminals are also located
in warm countries without the difficult winter conditions (snow, ice,
low temperatures) that can complicate the operation of automated

The greatest equipment.

demand for Thirdly, equipment decision making is centralised and executed in layers

automation that integrate the actual equipment control, the management of the
fleet of machines at the terminal (Terminal Logistics System, TLS) and
solutions will the overall operations of the terminal (Terminal Operating System, TOS).

come from Finally, the manufacturing volumes for container handling machines
are significantly smaller than those of the automotive industry. This has
the existing made it possible to deliver automation projects one by one to various

manually customers without needing to wait for a global standardised solution.

operated Today, automation is generally accepted as the primary way of

terminals improving the competitiveness, predictability and safety of container

terminal operations. It is unlikely that any new terminals will be built for
around the traditional all manual operation, but the greatest demand for automation

world. solutions will come from the large number of existing manually operated
terminals around the world, as they continue to seek competitive
advantage in a high-pressured global business environment.

12
AUTONOMOUS CARS TODAY

At the time of writing, autonomous cars are still at the experimental

stage. Almost always, a human is in the driver's seat, even if the car
is driving autonomously. The level of automation is increasing feature
by feature, but despite fast progress in enabling technologies, we are
still a long way from globally available commercial autonomous cars.
The main challenge that technology developers are addressing is
enabling autonomous vehicles to deal with the endless variety of driving
situations and exceptions that can be encountered in traffic conditions
around the world. Ultimately, this problem will be solved by machine
learning models that are able to utilise vast pools of real-world training
data gathered around the world.

A major difference with container terminal automation is the global

operational area required of autonomous cars. Differences in national
legislation, driving culture and weather conditions (fog, rain, dust,
snow, sunshine etc.) pose massive challenges for the development of
self-driving vehicles. Many production cars already have technically
impressive driver assistance features that include various degrees
of driving automation, and several major car manufacturers and
technology companies have announced ambitious plans to bring fully
autonomous vehicles to market over the upcoming years.

Equipment manufacturers are progressing stepwise in adding

automation features to vehicles, with the key technical capabilities
under development including sensor fusion and machine learning for
localising the vehicle and mapping the environmental context around it.

A defining aspect of automated car development is that decisionmaking

and navigation capabilities will likely be decentralised in each individual
vehicle, at least for the foreseeable future. As a consequence, advances
in onboard sensoring and processing capacity will incrementally enable
new driver assistance features that evolve towards the ultimate goal of
truly autonomous cars.

LEGISLATION AND INDUSTRY REGULATION

A major challenge for the development and adoption of autonomous

road vehicles is that existing legislation and safety standards generally
do not address self-driving vehicles. As a result, questions of liability,
safety certification and applicable safety levels need to be reconsidered
for autonomous cars. Changes in legislation and/or industry standards
usually take several years to implement, but the speed of technical
development in the field has outpaced the capability of policymakers to
adjust the regulatory framework.
 The container In the United States, Federal Motor Vehicle Safety Standards outline the
requirements for minimum safety performance, objective testing and
handling
AUTOMATED CONTAINER TERMINALS

manufacturer self-certification. However, the federal level has no safety

industry standards for autonomous vehicles, so references to a driver can be

interpreted either as requirements towards an autonomous driver (AI), or
already works a human driver/steward overseeing the operation of the vehicle.
AND SELF-DRIVING CARS

within a The U.S. Department of Transportation Automated Vehicles Policy

tightly defined from late 2016 is not a rulemaking initiative but an agency guidance

framework document that outlines steps for the deployment of self-driving vehicles,
defined as highly automated vehicles (HAV). The document sets
of industrial out recommendations and best practices in several areas, including

standards and requirements for pre-deployment testing, a model for a future state-level
legislative framework, as well as guidance on the applicability of existing
occupational and new regulatory tools.

safety Meanwhile, various states have progressed with their own legislation to
legislation. facilitate the adoption of autonomous vehicle testing, and the majority of
states already provide some provision for self-driving vehicles on public
roads; however, these state-by-state laws differ on even basic concepts
such as how to define the "operator" of the vehicle. Most notably,
California, Arizona and Nevada have updated their laws to allow the
testing of autonomous vehicles without a human driver inside.

The EU faces a similar situation, in which member states are

forging ahead with national regulations in attendance of a union-
wide framework on autonomous cars. No well-defined acceptance
procedures exist for road vehicles at higher levels of automation, but the
topic is under active discussion in several countries, as manufacturers
and research organisations proceed with their development efforts.
In Finland, autonomous cars at any SAE automation level may be
temporarily tested in traffic subject to a test permit, provided that the car
has a human driver either inside or outside the vehicle.

By contrast with autonomous road vehicles, the container handling

industry already works within a tightly defined framework of industrial
standards and occupational safety legislation. However, the challenge
here is similar in that no safety standards or laws exist yet specifically
for container terminal automation, so applications need to be based on
interpretations of existing norms and regulations.

14
4. Technology considerations

CLOSED/CONTROLLED VS. OPEN/UNCONTROLLED ENVIRONMENTS

Container terminals are closed environments, which has made it

possible to deploy extensive automation at many terminals. The car
industry and mobility as a service providers are attempting to start with
a parallel approach, since it is easier to introduce commercial operation
for self-driving cars in limited areas and applications (e.g. shopping
centre valet parking or public transportation in a selected city), rather
than globally.

The major challenge that the developers of autonomous/self-driving

cars are tackling today is ensuring the safe and consistent operation
of the vehicles in the infinitely variable open environment of road traffic.
In controlled or semi-controlled environments (such as people mover
systems at airports or corporate campuses) self-driving vehicles
have been operating safely and reliably already for many years. The
complexities of open and uncontrolled environments will require vastly
more advanced capabilities, as manufacturers and system designers
need to address an endless range of factors ranging from differences
in traffic culture, national legislation, weather conditions and road
infrastructure.

Driving on a straight road is a relatively simple task to automate, but

self-driving cars will also need to cope with snow, darkness, animals of
various sizes, unpredictable and sometimes irrational people in traffic,
various road surfaces, and traffic conditions that range from a desert
highway to a rush-hour roundabout in Paris or the organised chaos of
traffic in Mumbai.

This infinite amount of exceptional situations is the main source of

complexity in enabling global fleets of fully autonomous cars. It seems
likely that the only way to make significant progress with these cases
is to harness existing cars and humans to automatically gather and/
or crowdsource data from a vast range of situations and feed this data
into machine learning applications to create models that can react
to such situations. Thus, it is only the current exponential growth of
technology (and especially AI and machine learning) that offers any
reasonable chance of making autonomous cars a reality in the next few
decades.

At container terminals (ports, intermodal terminals or industrial logistics

centres), the main challenges of automation are, firstly, continuously
increasing the reliability and performance of load handling and,
 secondly, the fluid collaboration between load handling equipment,
other vehicles (trucks, trains and container vessels) and especially
AUTOMATED CONTAINER TERMINALS

people (operators, service personnel).

Industrial applications also have different safety requirements than

commercial vehicles. Container terminal automation takes place in
AND SELF-DRIVING CARS

a closed area with safe access control systems built to well-defined

industrial safety standards, though development is moving towards
operations in which automated and manual machines can work
alongside humans. Conversely, the car industry will always need to
operate in so-called mixed mode as conventional vehicles, smart cars
and people move together in open traffic.

Paralleling the autonomous vehicle industry, terminal automation

also meets its greatest design challenges at the interfaces between
humans and automated equipment, such as when loading road trucks
by automated RTG cranes. In such an environment, there are still
numerous variables and exceptional cases that need to be tackled
one by one, but the magnitude of the challenges is limited compared
Autonomous to the uncontrolled environment of a global fleet of autonomous cars.

vehicles rely At terminals in which the container handling equipment does not need
to physically interact with people, the degree of automation is already
on a wide several levels ahead of sites that must integrate people with automated

range of machines.

sensoring SENSORING AND PERCEPTION

solutions, with Autonomous vehicles rely on a wide range of sensoring solutions, and
development this area is already relatively advanced, with development continuing at

continuing at a fast pace. The solutions are based on the concept of sensor fusion,
combining inputs from numerous systems from inertial measurement
a fast pace. units, cameras and GNSS devices (Global Navigation Satellite Systems,
including GPS) to radar, ultrasonic and Light Detection and Ranging
(LIDAR) sensors. The vast amount of data generated by these systems
is then handled with artificial intelligence (AI) such as machine learning /
deep learning methodologies to enable the autonomous solution.

In the development of autonomous cars, the primary sensoring

technology varies from manufacturer to manufacturer. For example,
Tesla focuses mainly on camera-based systems while Waymo depends
more on LIDAR technology. At automated container terminals, the need
for advanced sensoring also extends from vehicle movement to the
handling of the actual containers, which can utilise various technologies
such as laser scanners.
Irrespective of the technical solutions used, both fields share the
common goal of enabling the solution to build a sufficiently detailed
real-time picture of the environment in order to ensure safe and efficient

16
 operation in any expected conditions. Thus, it is useful to extend the
concept of sensoring to the wider concept of machine perception in
both contexts.

As a high-volume mass market, the car industry is developing sensors,

actuators and cameras at a remarkable pace to boost the development
of autonomous vehicle systems. The major auto manufacturers are
investing heavily in this field, typically depending on an ecosystem
of specialised hardware and software subcontractors for the huge
range of subsystems required. Container terminal automation also
depends on the sensor fusion of data from satellites, radar, magnets,
transponders, cameras and laser scanners, with the exact combination
selected on a case-by-case basis. By contrast with the car industry,
these systems are usually built from safety-rated products designed
specifically for industrial purposes, instead of high-volume bulk
components.

The rapidly developing mass-market sensor technology in the car

industry will also boost innovation in industrial applications, while
simultaneously lowering component prices. Container terminals
therefore have the possibility of adapting these innovations very quickly
to their own sensing and perception systems, so this area holds the
potential for fast improvement already in the short and medium-term
future.

Sensor fusion in autonomous cars

Source: Society of Automotive Engineers

Surround View

Spot
Traffic Sign Detection
Recognition
Cross
Traffic
Alert
Adaptive Emergency Braking
Cruise Pedestrian Detection Park Park Park Assistance/
Control Collision Avoidance Assist Assist Surround View
Rear
Collision
Warning
Lane Departure
Warning

Surround View

Long-Range Radar
UDAR
Camera
Short-/Medium-Range Radar
Ultrasound
 MAPPING AND LOCALISATION
AUTOMATED CONTAINER TERMINALS

Another area of rapid development in the car industry is creating and

maintaining the highly detailed, dynamic 3D HD maps required by
autonomous vehicles. Currently technology companies have large fleets
of cars scanning the global roads continuously, but in the future new on-
AND SELF-DRIVING CARS

vehicle sensors will enable the increasing use of crowdsourcing for the
distributed collection of dynamically changing map data.

The ETSI (European Telecommunications Standards Institute) has

defined an LDM (Local Dynamic Map) framework for managing vehicle
sensor and map data, which has subsequently been standardised as
ISO standard 18750. Local Dynamic Map information encompasses
several levels from permanent and transient static data (map features,
roadside infrastructure) to transient dynamic data (e.g. congestion and
traffic signal phase) and highly dynamic real-time data such as vehicles
and detected pedestrians.

In current terminal automation solutions, map data is relatively static,

with a possibility for updates through various configuration actions.
Adopting dynamic HD-level mapping technologies holds the promise for
great improvement at future container terminals. In addition to changes
in terminal infrastructure, such dynamic 3D maps could include data
such as the position of people or service vehicles moving in the area, or
even holes in the asphalt that should be avoided by equipment. Even
though the terminal infrastructure design remains as the basis for the
environment map, it is possible to adapt features of dynamic maps from
smart/autonomous traffic to enable more dynamic changes in terminal
layout maps.

CONTROL

Independently operating self-driving vehicles – and even today's

cars with advanced driver assistance features (e.g. Tesla) can rightly
be described as computers on wheels, instead of cars that include
computers. Alongside sensoring/perception technology, the key
development area today is the processing required for the vehicle to
safely and reliably make sense of its environment and respond optimally
to the infinitely variable situations that can be encountered in different
traffic and weather conditions.

Arguably the single most important long-term enabler for autonomous

road vehicles will be AI, which is the core technology required for
accurate real-time situation analysis and safe decision making in
complex open traffic environments. Resultantly, a key challenge for
autonomous vehicle developers is managing the massive system
complexity required by high-level sensoring and AI deployments on
individual vehicles.

18
By contrast, automated container terminals already utilise a range
of highly reliable, field-proven solutions for equipment guidance,
navigation and control. Positioning/navigation technologies range from
infrastructure-based systems such as magnetic markers and radar
beacons to various on-equipment solutions. However, automated
container terminals operate with a different set of design requirements,
as the equipment is, by default, always controlled centrally and the
scope of operations is limited to the terminal.

Despite these differences in the underlying concept, the current rapid

development of AI for autonomous vehicle solutions is also a major
opportunity for the container handling industry, as the exponential
improvement and affordable availability of both on-vehicle and server/
cloud-based AI processing capacity will open up new possibilities for
centralised automated solutions in the container terminal environment.
As the technology for high-level AI becomes accessible and increasingly
commoditised, the cost or availability of raw processing power is
no longer a design constraint. Instead, the challenge becomes one
of identifying successful solutions in other fields and applying them
creatively to the specific requirements of the logistics industry.

Algorithms vs. machine learning

In any automated or autonomous system, some control and decision
making is always possible with algorithms created by software
developers. However, when the set of required rules and potential
exceptional cases gets large enough, the only possibility will be to
utilise various kinds of machine learning. Rather than hard coding
system behaviour into fixed rules, machine learning allows developers
to focus on gathering a lot of learning data and creating models based
on it. Currently there are several multilayer deep learning technologies
available, and new ones coming in the future, so one needs to
have the competence to select the right method for each problem.
A combination of fixed algorithms and various machine learning
techniques is the standard approach for complex problem sets such as
those encountered in autonomous cars and vehicles.

At the highest level, the key difference between self-driving cars and
automated container terminals is that in current terminal automation,
most of the control and decision making is based on pre-programmed
software-coded algorithms, though in the future machine learning
techniques will increasingly be used to support and improve control and
decision making. By contrast, in self-driving cars, it is impossible to get
far with pre-coded algorithms, so machine learning is a key success
factor both now and in the future.
 One should also remember that AI is already in widespread use in many
small applications, both in passenger cars and terminal automation. In
AUTOMATED CONTAINER TERMINALS

vehicles, AI is used e.g. for automatic traffic sign recognition, detection

of animals in order to avoid accidents, and automatic lane monitoring.
In container handling, successful production deployments already
include solutions such as optical character recognition for container
AND SELF-DRIVING CARS

identification, license plate recognition at automatic truck kiosks, and

automatic damage detection for containers.

Infrastructure
Infrastructure for self-driving cars is rarely given much attention in public
discussion, and is often viewed almost as a "nice to have" add-on. The
current focus of self-driving cars is to make them standalone products
that can drive autonomously, and manufacturers cannot count on a
global infrastructure being available solve the problems associated with
car operation and control. As a result, smart traffic infrastructure will
probably add value to autonomous cars on a case-by-case basis. For
example, when self-driving cars are used in predefined applications in
constrained areas (e.g. automatic “valet parking” at shopping centres),
Artificial the location can be equipped with the necessary infrastructure to enable

intelligence for such a mobility service. Even though various kinds of sensors in traffic
lights, traffic signs, roads etc. may add value beyond the standalone
autonomous decisionmaking in the cars, the global self-driving car industry does not

vehicle expect infrastructure-based solutions to solve its major design problems.

solutions is By contrast, in the limited area of a container terminal, it is

also a major straightforward to build the local infrastructure to enable automated

applications. Whereas autonomous cars are basically designed to
opportunity for operate on their own in open environments, terminal automation can be

the container designed with a specific infrastructure in which most of the obstacles
except people and vehicles are known and defined beforehand.
handling
industry. Communications
Vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) machine
communication – also known as Connected Driving – hold tremendous
potential for boosting the safety and efficiency of both manually driven
and automated cars. In the future, cars and infrastructure may start
sharing relevant information with each other wirelessly, such as the
vehicle's own state, speed, location, traffic light status, approaching
emergency vehicles, slow vehicle warnings, platooning coordination etc.

In order to reach this type of functionality, the vehicles and infrastructure

require several levels of static and dynamically updated map data. The
standards and base technology for Connected Driving already exist in
the previously mentioned ISO standard 18750, so the capability could
theoretically be introduced to new cars, trucks, buses and motorcycles
at any time. Despite this, the industry has until now been slow to
embrace the concept. Since both manually driven and automated

20
vehicles stand to benefit from Connected Driving, its eventual impact
may turn out to be larger than that of automatic driving alone.

In connected cars, continuous high-speed data connectivity is required

for real time map updates, as well as to enable smart features that
assist drivers by enabling faster reactions than those of a human
driver. This kind of smart traffic communication is one of the basic
building blocks of autonomous driving. 5G and modern mesh/swarm
communications will improve the connectivity of both V2V and V2I
communication, thus leading to advances in safety and traffic efficiency.

At container terminals, all automated operations are, by default, based

on real-time connectivity, so future technology improvements will also
add value to terminal applications. At terminals, cables and fibre-optic
connections can be used for certain machines, but many operations
depend on reliable high-speed wireless communications. Wifi can
currently be used for some applications, but 5G data communications
are expected to provide significant new possibilities for connectivity at
all automation levels. In the future, machines may also communicate
with each other to a greater extent.

MAINTENANCE ASPECTS

Smart maintenance – in which systems are equipped with advanced

monitoring, diagnostic and telemetry features – will be essential for all
kinds of autonomous vehicles. Sensors and systems on the vehicle will
be collecting condition data that will be used for smart decisionmaking
about the most optimal time of maintenance.

At automated container terminals, the focus of maintenance shifts

from ad-hoc repairs to preventative maintenance. It is notable that
even as automation increases the predictability and availability of
container handling equipment, it also requires a more stringent service
programme that focuses on preventative maintenance tasks and
continuous monitoring of equipment condition. At manual terminals,
human operators can often compensate for small deficiencies or
malfunctions in equipment, but automated machines always need to be
at full performance for the system to deliver its expected potential.

Automated container terminals require more – not less – attention to

maintenance, to ensure that machines are not broken and un­available,
because there is no human driver or operator there to notice basic
faults before they develop into more serious malfunctions. It is
reasonable to assume the smart and predictive maintenance
programs will also be essential for self-driving cars. On the other hand,
automatisation goes hand in hand with the ongoing trend towards elec­
trification. A detailed discussion of electric cars is beyond the scope
 of this paper, but it should be noted that in both cars and container
handling equipment, electric drivelines feature a completely different
AUTOMATED CONTAINER TERMINALS

maintenance paradigm, potentially offering higher reliability and lower

lifecycle costs. However, the level of technical competence required to
service and maintain automated/electric equipment is also very different
from traditional vehicles or machines.
AND SELF-DRIVING CARS

SYSTEM APPROACHES: AUTONOMOUS VEHICLES VS. CENTRALISED

DECISION MAKING

Autonomous container handling equipment has always been – and is

likely to remain – centrally controlled, even as the intelligence of the
equipment increases rapidly. Though the ultimate goal is an automated
system that will be able to handle any conceivable situation, remote
control or teleoperation will still be needed for exception handling far
into the foreseeable future.

The development of autonomous cars is currently focused almost

exclusively on vehicles that distribute the decision making and
processing to each individual vehicle, at least if the discussion is limited
to private cars and new mobility services instead of a larger context
of future public transportation infrastructures. Whether autonomous
cars can actually succeed in fulfilling many of the expectations set for
them (reduced traffic congestion, improved sustainability, less need
for parking in city centres etc.) without shifting the underlying design
approach to a centrally managed system remains a space for some
debate. However, such a major paradigm shift on a global scale seems
highly unlikely, as long as individual transport and freedom of mobility
remain the prevailing core values for consumers/road users and vehicle
manufacturers.

EXPONENTIAL TECHNOLOGY GROWTH

Exponential technology growth has been with us for decades, from

the processing power of integrated circuits to memory capacity and
communications technology. A high-capacity hard disk in 1956 might
have been 5 MB in capacity; today, our portable drives hold Terabytes
of data for a tiny fraction of the cost. The wireless data speed of GSM
in the 1990s was 9.6 kbps, while 5G will reach speeds of 10 Gbps with
sub-millisecond radio latency in the next decade. As impressive as these
figures are, they don't even begin to address the potential of how various
new applications can accelerate the development of other industries.

As far as technology is concerned, many commentators have

posited that we have already moved well into the "second half of
the chessboard" – the point where the speed of exponential change
becomes difficult to understand. This has particular relevance to the

22
Exponential accelerating onset of digitalisation and how it will affect global maritime
container logistics. Exponential technology growth in computing power,
development memory, data communications and cameras will drive the adoption

means of new solutions at speeds that may be impossible to estimate

beforehand.
inevitable
business The exponential growth of digital technology – and especially AI /
machine learning – is currently the key enabler of autonomous vehicle
change. The development. However, this development also brings with it much

winners will be wider implications. Exponential development means inevitable business

change, and the winners will be those that are the fastest to adapt to
those that are the ongoing change in the world.

the fastest to In a world of exponential development, well-run established companies

adapt to are actually the ones that are most at risk as new entrants shake up

the ongoing the business models. Most often, the biggest change happens in the
creative crossing of existing industries. So the question becomes, does
change in the the container handling industry dare to challenge itself and utilise the

world. advances of other fields (including autonomous car development) or

does it choose to wait to be disrupted? And what is the future of the
car manufacturers that have ruled their markets for a hundred years or
more? Can they cope with challenges from new service providers, or
will they themselves shift towards MaaS business models?
 5. Safety
AUTOMATED CONTAINER TERMINALS

FUNCTIONAL SAFETY

Autonomous cars and automated container handling equipment need to

AND SELF-DRIVING CARS

conform to strict existing safety requirements as well as highly demanding

safety expectations from stakeholders within and outside their own industries.

A key concept for system designers is that of functional safety, which refers to
the automated, predictable and correct response of a system to its inputs, or to
the system having a fail-safe design in case of malfunction. In road vehicles, the
core risk classification scheme is the Automotive Safety Integrity Level (ASIL),
which is adapted from more general industrial safety level standards. Defining
the ASIL of a given system or product involves an analysis of the severity,
exposure and controllability of the hazards in a vehicle operating scenario. The
desired safety goal for the hazard then defines the required ASIL.

ASILs are classified to four levels, with ASIL D dictating the most stringent
integrity requirements and ASIL A the least demanding. The development and
certification of software and hardware-intensive advanced driver assistance
solutions to ASIL D has proven to be a challenge for equipment manufacturers;
as system complexity in autonomous vehicles increases, this challenge is likely
to remain.

STANDARDISATION AND INDUSTRY NORMS

Automotive safety standards set out a comprehensive and detailed framework

that manufacturers must follow for commercially available road vehicles. The
difficulty is that no industry-wide standards exist yet for autonomous cars.
Even as several states in the US forge ahead with legislation that approves
self-driving cars without a safety driver behind the wheel (though usually with a
remote operator and/or special permits), the industry must do its best to adapt
current systems to existing standards.

The key safety standard for road vehicles is ISO standard 26262, titled "Road
vehicles - Functional safety", which defines the above-mentioned ASILs. An
ASIL is calculated as Severity x (Exposure x Controllability), thus taking into
account the possible consequences and likelihood of a system failure, as well
as whether a driver would be able to take action to prevent injury.

ISO 26262 is adapted from the IEC 61508 functional safety standard that
provides a more general framework for designing and deploying automatic
protection systems. IEC 61508 utilises a risk class matrix that rates various

24
Manufacturers hazards on the likelihood of their occurrence (from Frequent to
Incredible) as well as their consequences (from Catastrophic to
and system Negligible). Though similar in overall concept, there is no direct mapping

designers between the SILs defined in IEC 61508 and the ASILs of ISO 26262,
with the latter involving a more complex risk mapping with three
must embrace separate dimensions (severity, exposure and controllability).

a philosophy In the field of industry, a key safety standard is ISO 13849, "Safety of
of humanised machinery – Safety-related parts of control systems", which also refers

automation. to IEC 61508. ISO 13849 provides safety requirements and guidance
on the principles for the design and integration of safety-related parts
of control systems, including the design of software. For these parts, it
specifies characteristics that include the performance level required for
carrying out safety functions. The standard can be applied to systems
with high demand and continuous operation, irrespective of the type of
technology and energy used.

HUMAN FACTORS

In both autonomous cars and terminal automation, it is easy to get

fixated on the questions of technology, and forget that in both fields,
manufacturers are designing solutions for people, in a stepwise
approach to help them in their work, business and free time. To gain
wider adoption, any solution must ultimately address the needs,
goals and values not only of its immediate users, but also of other
stakeholders and society as a whole.

In all dealings with automated and autonomous machinery, the most

complex issues always arise at the interfaces where people need to
work with and/or share space with automated equipment. Beyond the
obvious requirements for safety, the demands extend much deeper into
questions of how we interact with our machines. Whether designing
self-driving vehicles or automated container handling equipment,
manufacturers and system designers would do well to embrace a
philosophy of humanised automation, all the way from enabling the safe
and seamless operation of machines and people in the same area, to
top-notch user experiences that ensure our systems serve our needs,
and not vice versa.

SECURITY

In addition to occupational and consumer safety aspects, automated

equipment, self-driving cars and any type of autonomous system also
raise a wide range of new questions related to security. New threats
 will inevitably surface, and new solutions will need to be found to
address them.
AUTOMATED CONTAINER TERMINALS

Potential threat scenarios may range from intentional unavailability

of systems, to safety hazards created on purpose. How do we stop
malicious actors from hacking an autonomous car in order to kill a
AND SELF-DRIVING CARS

pedestrian on purpose – or a terrorist from using one to deliver a bomb?

What completely unexpected threats might be made possible by the
new technology being deployed at the cutting edge of development?
Especially in the open, global operating environment expected of
autonomous cars, the security discourse is likely to become extremely
complex as the deployment of self-driving vehicles progresses further.

HIGH-LEVEL SOCIETAL QUESTIONS

Even though we are still awaiting a major breakthrough in the adoption

of autonomous road vehicles, the rapid development in the field
has already called for a wide-ranging debate on how to define the
responsibilities, liabilities and accepted risk levels of self-driving cars.
When algorithms and “smart” machines perform decision making in
life-or-death situations, the question of who is ultimately to be held
accountable becomes extremely complex and laden with legal, political
and moral dimensions.

In a few highly publicised incidents over the last years, autonomously

driving cars or advanced driver assistance systems have been
involved in road accidents resulting in loss of life. Paradoxically, the
further autonomous vehicle control systems advance and the safer
they become, the harder it will be to assign responsibility on the
occasions that they do fail. Can unequivocal blame be laid on the car
manufacturer, software designer, or some other party further in the
system development chain? Or does the responsibility for the safety of
the vehicle and its occupants ultimately always rest with a human driver/
steward who should theoretically be ready to step in if the autonomous
functions of the car fail to cope with a given situation – even if at
highway speeds there is no realistic chance of a human being able to
react and take control fast enough?

As a society, we need to make difficult decisions on our expectations

and demands of autonomous vehicles, and how we measure the
fulfilment of these demands. Should we expect self-driving cars to be as
safe, or safer than human-driven automobiles per kilometre travelled?
Or safer than trains, or commercial air traffic? How do we deal with an
autonomous car accident that a human driver couldn't reasonably have
averted? When loss of life occurs, can an autonomous car manufacturer
defend itself by stating that even after the accident, the car in question
has already been safer on average than traditional vehicles?

26
For automated container handling equipment operating in the closed
environment of a container terminal, the moral issues are not quite as
daunting, but manufacturers and software developers still need to deal
with the formidable task of ensuring the safety of their systems, even as
the complexity of automation deployments increases. As development
is extended to ecosystems involving multiple parties with specialist
knowledge, it becomes an increasingly challenging task to not only
design but also to certify systems to whatever safety level is required.

For evaluating safety levels and identifying areas for further

improvement, modelling, simulation and real-world data are essential.
In terminal automation, the ultimate responsibility for accidents is often
focused on the equipment/system manufacturer, which can create
pressure for over-engineering or hinder the adoption of new solutions.
Ultimately, we want to create solutions that are smartly safe – i.e.
designed for human needs, rather than over-engineered based on other
aspects such as legislation or the opinions of engineers.

In both autonomous cars and terminal automation, solution

development relies on vast amounts of data being gathered on
the autonomous/automated machines, as well as their operating
environments and conditions. AI guidance and collision avoidance
systems require massive real-world datasets to train their algorithms,
and autonomous car manufacturers make a point of advertising
the millions of kilometres their vehicles have driven without incident.
To further accelerate the development of safety in autonomous
vehicle design, an open, industry-wide culture of proactively sharing
safety-related data would enable all manufacturers – and ultimately
society as a whole – to benefit from the experiences of others.
However, competitive pressures and the proprietary interests of each
manufacturer may render such an ideal scenario unfeasible in practice.

Likewise, a global pool of source data for machine learning could help
developers solve the vast amount of exceptional cases faster in order to
create a significant number of fully automated terminals all around the
world. For example, gathering data from twistlock operations, container
moves, truck alignment, routing, and navigation operations would
enable continuous self-learning based on actual operational data.
Finally, real-world data on safety hazards and near-miss cases would
enable improvement of safety throughout the industry.
 Future prospects
AUTOMATED CONTAINER TERMINALS

TOWARDS THE AUTONOMOUS WORLD

AND SELF-DRIVING CARS

At the time of writing, the development of or autonomous cars is

proceeding at an impressive pace, with many of the world's leading
car manufacturers and technology companies making huge research
and development investments in the field. For example, Volvo has
announced that it intends to bring its first unsupervised autonomous
vehicles to the market by 2021. The stepwise adoption of more
advanced driver assistance features in commercial cars will eventually
lead the way to the ultimate goal of an autonomous car that meets the
technical, safety, price and performance demands of the mass market.

How much of – and when – this projection actually becomes a

reality remains to be seen; manufacturers are happy to provide quite
ambiguous estimates of when their autonomous cars will arrive, or even
what the term 'autonomous' specifically means in the context of a press
release. It is well understood that building a vehicle that can consistently
navigate suburban streets in sunny Arizona is quite different from
equipping a vehicle to do the same on a rural road in northern Sweden,
while safely dealing with snow, fog and reindeer crossing the road.

By contrast, the container handling industry is already years or decades

into the successful production use of automated solutions. Automation
is generally accepted as the primary way for terminals to improve their
operations over the upcoming years, and as solutions develop, the
benefits of incrementally deployed process and equipment automation
will become increasingly accessible to terminals of all sizes.
The greatest
value of However, the greatest value of terminal automation will come when it is
eventually combined with the new digitalised and connected logistics
terminal chains that are rapidly taking shape around the world.
automation will It remains to be seen how much autonomous and automated operation
will change the roles and value creation in the global logistics business,
come when it but a fully digitalised, transparent supply chain holds great promise
is eventually for not only the container shipping business but also the whole global
economy.
combined with
the new Eventually, autonomous container handling equipment, cars, trucks,
vessels and trains will all work smoothly together, transporting
digitalised and containers between automated terminals. In the future, even the
connected containers may become smarter and more connected thanks to
advances in the Internet of Things (IoT). Simultaneously, the surrounding
logistics infrastructure will also be connected with every car, machine, container
chains. and device, enabling cybersecurity threats to be tackled resiliently.

28
Software In this autonomous world, we will see fewer human-coded algorithms,
as developers focus on self-learning models and the harnessing
reliability of big data from various sources including crowdsourcing. This will

may prove aid systems in coping with the never-ending exceptional situations
encountered in the real world. System intelligence will move from the
to be one cloud to the edge, making cars and machines more intelligent and

of the most enabling them to make more independent decisions. Standardisation

and openness will increase, and ports, intermodal terminals and other
severe limiting industry facilities will adopt and integrate mass-market technology from

factors for the auto industry into terminal automation.

autonomous DEVELOPMENT DRIVERS VS. INHIBITING FACTORS

car adoption.
When trying to predict the future of autonomous vehicles from today's
vantage point, it is worth considering some of the potential factors that
may inhibit their wider adoption. Many of the utopian scenarios with
which autonomous cars are currently marketed may also need some
re-evaluation when subjected to real-world constraints. By contrast,
the container handling industry is in a relatively advantageous position
in that solution development is mostly contained to the limited and
controlled environment of the terminal, so many of these challenges are
not felt as acutely.

Software reliability may well prove to be one of the most severe

limiting factors for autonomous car adoption. Developers are faced
with the formidable task of designing vehicles whose AI capabilities
will eventually allow them to function reliably, safely and predictably in
chaotic inner-city environments or rough rural conditions. Susceptibility
of the car's sensing and navigation systems to different types of
weather (such as snow), let alone deliberate interference such as
jamming or spoofing will remain a concern. The avoidance of large
animals requires instant recognition and tracking, and software that
is optimised for caribou, deer, and elk may prove ineffective with
kangaroos.

Autonomous cars will also require very high-quality specialised maps

to operate properly. If and when these maps are inaccurate or out of
date, the vehicles need to be able to fall back on reasonable failsafe
behaviours. Competition or interference in the radio spectrum required
by the car's communication systems may pose problems, and current
road infrastructure may need extensive, costly modifications for
autonomous cars to function optimally.
 The total cost of ownership (purchase, maintenance, repair and
insurance) of autonomous vehicles is still unknown, even if novel cost-
AUTOMATED CONTAINER TERMINALS

sharing models may some day change the basic paradigm of private
car ownership. Finally, differences of opinion on how governments and
lawmakers should respond to the development of autonomous cars
may cause delays in their acceptance for the road. Irrespective of which
AND SELF-DRIVING CARS

path is chosen, the regulatory and standardisation framework will be

one of heated discussion and intense debate.

Autonomous cars have been predicted to have a huge impact on

almost every area of business. Disruptive change is taken as a
given, even if the specific predictions of what this change involves
are sometimes in direct contradiction. Often, the public debate
conflates arguments about autonomous vehicles with those regarding
electrification or new models for vehicle sharing. On one hand, mobility
as a service (provided by autonomous vehicles) is expected to decrease
private ownership of cars; on the other, autonomous cars may attract
new owners who have previously felt uncomfortable with the idea of
driving. Mobility as a service is seen as a way to introduce self-driving
cars more rapidly into markets, because the operation can be limited
rather than global.

Autonomous cars are predicted to decrease traffic congestion, vehicle

emissions and the need for parking in city centres; at the same time,
it remains difficult to see how this may come to pass if huge fleets
of roving autonomous vehicles are required to serve the on-demand
mobility needs of the working population. It is equally easy to conceive
of a situation in which the adoption of self-driving vehicles actually
increases traffic in city centres. If an autonomous car can roam
the streets and park on its own after dropping off the driver at the
destination – as currently advertised by e.g. Volvo – does the car owner
have any incentive to park the car at all? Why not let it drive around
for the day, and have it pick you up after work – or send it back home
to the suburbs for parking, effectively doubling the daily commute
distance?

If the promises of car manufacturers come true, autonomous cars

have the potential to hugely improve the safety of road traffic, but even
this positive development may have unexpected side effects. How
will our healthcare system deal with a significantly smaller number of
organ donors from traffic accident deaths? Will our pizza taxis become
automated for last-mile delivery? If self-driving cars become mobile
hotels or replacements for medium-haul flights, what does it mean for
the hotel and airline businesses?

30
KEY OPPORTUNITY: SELF-DRIVING CARS BRING NEW CAPABILITIES TO
THE TERMINAL INDUSTRY

The rapid mass-market development of supporting technologies for

autonomous cars (AI, sensoring, guidance technology etc.) is the major
opportunity from which container terminal automation providers can
benefit today and in the near future. Instead of needing to develop
proprietary technologies from scratch, or systems based on expensive
industrial components, the industry can adapt and reuse many of these
new solutions for its own specific applications.

This opportunity is particularly powerful if it can be combined with

open standards and interfaces that allow for the creation of wider
business and technology ecosystems beyond the borders of individual
companies. Within the terminal industry, Kalmar has taken the first
steps to facilitate this kind of development with its Kalmar Key initiative;
whether automotive companies and technology providers are able to
converge on a workable set of open standards to benefit the growth of
self-driving cars will likely be a key success factor for the entire wider
industry in the years to come.

The future of self-driving cars:

• Continued development of new driver assist features
• MaaS services with autonomous cars, first in limited areas
• Development of support infrastructure
• Communications network development
• HD maps and crowdsourcing
• Eventual global deployment of autonomous cars, perhaps area by
area
 TIMELINE
AUTOMATED CONTAINER TERMINALS
AND SELF-DRIVING CARS

Step 4

Step 3 Global operations

(exact timeframe unknown,
probably decades)
Step 2 Autonomous
control approved
for limited areas
Step 1 Mapping evolves and services
relatively quickly

Sensoring part
solved first

´THE FUTURE OF CONTAINER TERMINALS

• More mixed mode traffic with manual and automated vehicles
together
• Sensor fusion and AI based container handling, obstacle detection
and control
• AI helps container handling operations, routing, path and storage
planning, safety, ship & truck operations
• Dynamic maps at terminals
• Wireless communications evolution fulfilling even the most
demanding requirements and enabling flexibility for retrofit terminals
• Sensors, cloud services, computing power face exponential growth
enabling improvement with automation and digitalisation
• Eventually moving towards fully automatic and autonomous
logistics and transparent end-to-end logistics chain
• Centralised control remains even as intelligence moves from cloud
to the edge (machines and their sensors)
• Once regulations for self-driving cars are solved, same framework
may be adapted to terminals

32
Summary
Autonomous cars and the automation of container handling equipment
share many common drivers but differ markedly in several respects. In
both fields, development proceeds stepwise towards full automation,
but a global mass market (car manufacturing) can benefit from greater
economies of scale in technology development than a comparatively
specialised industry such as our own. The underlying differences in
basic technical approach (autonomous control of self-driving cars
vs. centralised control of terminal automation) will remain for the
foreseeable future.

The approach to system safety is also very different between the two
fields. The container terminal business needs to adhere to a strict set of
pre-existing industrial safety regulations, whereas a high-volume market
led by the world's largest major industrial and software companies
might simply forge ahead with the development of self-driving cars
and expect legislation to catch up. For better or for worse, the terminal
business does not have this option.

The general trend is towards increasing autonomy in AI systems, and

increased autonomy always introduces increased chances of error.
Problems related to AI safety are most likely to manifest in scenarios in
which the AI system exerts direct control over its physical and/or digital
environment without a human in the loop – for example, automated
industrial processes, self-driving cars or cleaning robots.

Container terminals were among the first leaders in automated work

machines, and they have been able to build extremely reliable and well-
performing solutions with simpler technology in the closed environment
of a container terminal. Now, terminals have the unique opportunity
of picking the best of the new technologies being developed by the
mass-market car industry, and harnessing them for their own purposes.
However, due to significantly different operative business logic, the key
challenge will be learning to adapt and customize these capabilities to
the port environment. As for what kinds of next-generation solutions
we are able to create, the only limits will be in our imagination and our
capability to apply these technologies in our own field.
 Sources and further
AUTOMATED CONTAINER TERMINALS

reading
1. Society of Automotive Engineers (SAE) automation levels
https://www.nhtsa.gov/technology-innovation/automated-vehicles-safety
AND SELF-DRIVING CARS

2. United States Department of Transport Automated Vehicles Policy

https://www.transportation.gov/AV/federal-automated-vehicles-policy-september-2016
3. ISO standard 26262-1:2011, Road Vehicles – Functional Safety
https://www.iso.org/standard/43464.html
https://en.wikipedia.org/wiki/ISO_26262
4. ISO standard 13849-1:2015, Safety of machinery – Safety-related parts of control
systems https://www.iso.org/standard/69883.html
5. IEC standard 61508, Functional Safety
https://www.iec.ch/functionalsafety/
https://en.wikipedia.org/wiki/IEC_61508
6. ISO standard 18750:2018, Intelligent transport systems
– Co-operative ITS – Local dynamic map
https://www.iso.org/standard/69433.html
7. MIT online lecture series 6.S094: Deep Learning for Self-Driving Cars
https://selfdrivingcars.mit.edu

34
AUTHORS

JARI HÄMÄLÄINEN PEKKA YLI-PAUNU PERTTI PEUSSA

Director, Terminal Automation, Director, Automation Research, Principal Scientist and Senior Project
Kalmar, (Dr. Tech.) has Kalmar, (Lic. Tech.) has Manager (IPMA C), VTT, (Lic. Tech.)
a background from the background from automation has joined VTT already in 1988 and
telecommunications and software research in VTT and from factory often works as a project manager
industries, with over 300 patents automation/robotics projects. in industrial contracts dealing with
in 40 global patent families helping Pekka has started his career in smart machines, environment
smartphone users in their daily Kalmar in 2003, and has worked perception, and automated driving,
business and pleasure. He has in various automation R&D from which he has about 25-years
been working for Kalmar since 2012 management roles. Currently, experience. He has also worked
and is currently leading Kalmar Pekka is focusing on automation as a research group leader and
software and automation systems applied research projects and research manager but now project
development, simulations lab and building technology strategy and management is his priority.
test yard automation operations. He roadmap for coming years.
is also project director of the Kalmar
AutoRTG product development.
 Making your every move count

Published by Kalmar, part of Cargotec. Copyright© Cargotec 2018.

All rights reserved. No part of this publication may be copied or
reproduced without permission of the copyright owner. The content
of this document is provided “as is”, without warranties of any kind
with regards to its accuracy or reliability and excluding all implied
warranties. We reserve the rights to make changes to any of the
items described in this document without prior notice. The content
of each service and availability of particular services may vary.

www.kalmarglobal.com