IEEE Robotics & Automation Magazine - December 2018 - 111

safety development pipeline, as presented by Takacs et al. [9].
Ethics and Social Questions
As the enabling technologies of selfdriving are being developed, society
demands an understanding of the
operation principles and consequences
of an increased number of self-driving
vehicles on the road. Although AI has
been a tool used for some years now in
diagnostics, forecasting, and decision
making for many professional fields
(including finance, advertising, and
economics), self- driving cars may
become the first safety-critical system
that will have a worldwide effect on
society. It is a challenging task to forecast what direct social impacts we are
going to experience once self-driving
vehicles become an integral part of our
lives; it is even more difficult to address
indirect consequences brought about
through such ideas as car sharing,
transforming our commute time to
more efficient working time, and
changing urban layouts [10].
Even though self-driving cars will
increase road safety and provide a significant decrease in road injuries (there
were 1.26 million reported fatalities in
2015), an intense ethical discussion on
the self-driving car's moral decision
making has been going on recently. This
will require an educated public discourse, sparked by such promising
initiatives as the Robotic and AI Governance Initiative [11], leading to an ethical code of conduct or robot manifesto
[12]. Many of these discussions can be
traced back to the classic trolley problem, where an ethical decision has to be
made concerning whether to switch
tracks of a runaway trolley to sacrifice
one life in exchange for five other lives.
The Massachusetts Institute of Technology (MIT) projected this problem to the
decision making of self-driving cars.
The MIT Moral Machine [21] generates moral dilemmas and collects
information on the choices participating volunteers make between destructive outcomes. The generality of these
problems can be illustrated by the fact
that a variant of the ethics guidelines
used in medicine-deontology and

utilitarianism-can be observed in the
case of self-driving cars as well, with
either a society-centered or individual-centered approach being taken.
Although some of today's car manufacturers have already taken a stand on
one side or the other (e.g., Mercedes-
Benz will always prioritize to protect
its passengers), currently there is no
technological evidence that AI in selfdriving vehicles would be forced to
make such decisions, whether inherently based on software architecture or
learned via training methods. This
allows engineers to approach the problem of self-driving primarily from the
functionality side, which will ultimately
lead to better road safety and more efficient mobility.
The need for safety has been
broadly recognized by the engineering
community, and thus IEEE called to
action the Global Initiative on Ethics
of Autonomous and Intelligent Systems, which aims to support safe
autonomous system development via
ethical standards (the P7000 standard
series) [13]. To overcome the compatibility challenges involved in applying
traditional processes to the higher
level of autonomy, automotive safety is
on a path to integration into the technology development workflow from
the concept level all the way to testing
and validation.
Standardization and Clearance
Because the whole field is emerging and
evolving rapidly, autonomous vehicle
technology presents a huge challenge
to standardization bodies and legal
authorities. The objective assessment
and quantification of self-driving algorithms and systems for safety is now the
focus of numerous standards developing organizations (SDOs):
● ISO Technical Committee (TC) 022
(Road Vehicles); ISO TC 204 (Intelligent Transport Systems)
● ISO AWI PAS 21448; Road Vehicles-
Safety of the Intended Functionality
(Working Group 08 under the TC
022/Subcommittee 32)
● IEEE 1609.2a-2017, IEEE Standard
for Wireless Access in Vehicular
Environments-Security Services for

Applications and Management
Messages
● ANSI ITSDF B56.5-2012, Safety
Standard for Driverless, Automatic
Guided Industrial Vehicles and Automated Functions of Manned Industrial Vehicles
● ASTM Committee F45 on Driverless
Automatic Guided Industrial Vehicles.
Some of these activities are connected to the ongoing general robot
standardization activities of the large
SDOs, such as ISO TC299 [14]. Nevertheless, these are all technical standards, aimed at
addressing the
technical issues
Self-driving cars
listed previously.
may become the first
Another major
area for L4 and
safety-critical system
L5 validation is
that will have a
liability and ethical responsiworldwide effect
bility. This has
on society.
much in common with the
field of medical
robotics, where the increasing level of
autonomous functions raised similar
concerns [15].
In current serial-production cars,
only ADAS functionalities can be found.
From the viewpoint of safety, this defines
cooperative driving, i.e., the human and
the vehicle handle the driving task
together. On the other hand, L3 systems
may self-drive in place of the human
driver but still not be able to handle complex emergency situations. According to
current automotive safety standards, L3
systems must be able to recognize emergency situations where the intended
functionality cannot be carried out, shifting the full responsibility back to the
human driver. However, this handover
process is not yet standardized [8].
Conclusion-Why
Should You Care?
The research and development community quickly took on the new challenges presented by self-driving
technologies; now, the automotive
industry is gradually turning toward
autonomous solutions as well. It is big
business, as the recent investment

december 2018

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

111
IEEE Robotics & Automation Magazine - December 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2018
