IEEE Robotics & Automation Magazine - March 2022 - 95

their environmental lifecycle. Material
extraction and assembly, energy consumption
when used, and waste generation
when damaged or outdated are
several such effects.
Ethical Issues
With their many sensors and their connectivity,
the deployment of robots
already raises privacy concerns. As seen
during the COVID-19 crisis, the protocols
used to handle data by technological
solutions must be carefully designed,
together with adequate communication
strategies [34]. Privacy is not the only
issue: for example, robotics used to
influence behaviors raise the issue of
autonomy, while surveillance robots
raise a democracy problem.
Safety and Critical Performance
A system with safety deficiencies or
which does not accomplish critical
tasks correctly has the potential to generate
significant negative impacts on
the SDGs.
Impact on Norms and Values
Some panelists raised the effects of the
wide adoption of technology on the
norms and values of a society. The panelists
feel that these impacts should also
be considered.
SDGs/Ethics Washing
An improved awareness of the public
may push companies and governments
to build a virtuous image with respect to
SDGs or ethics considerations when
deploying robotic technologies. Without
concrete and continuing evaluations
of such claims, a trend of AI ethics
washing, or SDGs washing, could
appear similar to green washing with
environmental issues. This could be
detrimental as it would reduce the trust
of the public toward such efforts.
Developing SDG-Relevant
Robotic Systems
In addition to the potential undesirable
impacts of a technological solution, its
actual usefulness when deployed is not
always up to expectations. A panelist
gave this example: during the COVID-19
pandemic, a survey identified more
than 232 prediction models developed
to diagnose the virus' propagation. In
total, only two of them were deemed
actually promising and usable [35],
while the situation is similarly grim
with the prediction models for the loss
of lives [36].
During the process of designing a
robotic system to contribute to the
SDGs, the panelists stressed the need to
evaluate the context-specific tradeoffs in
the long term of robotic intervention
compared to another solution, which
may not be technological. To do so, it is
important to start with the human
aspect of the problem and consider the
entire ecosystem with a systemic
approach. Only then it is possible to
identify where technology could help as
an actor in the system, and evaluate
whether it would create problems
somewhere else. To correctly achieve
this assessment, interdisciplinary expertise
is needed as well as exchanges with
different stakeholders.
In addition, when the process
includes a collaboration among robots
and humans, it is important to capitalize
on their complementary strengths.
Humans, for example, are stronger with
empathy, social skills, and complex analytical
thinking.
Governance Frameworks for
Robotics and the SDGs
The field of robotics and the closely
related field of AI, pose a special risk
because of their potential for highspeed
and large-scale changes, and
because of the unpredictability in case
of negative consequences [37]. To support
the development and deployment
of robotics in the direction of the SDGs
and to prevent its unethical uses or
undesirable impacts, organizations and
policymakers can adopt governance
frameworks [38], [39]. Governance
must be understood as more than compliance
to rules: the legislation is often
lagging, and the rules and ethical standards
are different across organizations
and around the world. It requires building
a strong culture around sustainability
and ethics. In the following section,
we identify what is needed for these
frameworks to be effective.
Governance Over the
Technology Lifecycle
To address all impacts of the deployment
of a robotic technology, a governance
framework should be built for
the full lifecycle of these products:
design, development, production, and
deployment, also adding usage and
potential AI drift to the application
[40]. The community, together with the
authorities, shall develop measures of
transparency and traceability that will
be common across the different organizations
and stakeholders taking part in
the use of the technology.
Shared Responsibility,
Traceability, and Transparency
The framework should identify the
main actors and ascribe a shared
responsibility across the whole lifecycle,
including people involved directly or
indirectly. An example that was provided
by one of the panelists is the use of
facial-recognition technology. Although
it may have a large value for public safety,
facial-recognition technology raises
privacy issues and inequity concerns
when there is bias in the data used to
train the model. To share the responsibility
with respect to these issues among
those who develop the technology,
those who choose it, and those who
deploy and use it, transparency in the
different processes and traceability of
the data are critical.
Governance frameworks must therefore
demand a sufficient level of transparency
and traceability from companies
and organizations. In this context, it is
essential to implement interpretable
models [41] so as to ensure accountability
and transparency of the applications
where they are deployed.
Global and Regional
Frameworks
Although a global governance framework
may seem appealing, the panelists
argued that communities have different
understandings and values and would
be better served with local frameworks.
International regulations may be used
as backbones to be instantiated nationally
or regionally by states or organizations.
Limited measures could also be
MARCH 2022 * IEEE ROBOTICS & AUTOMATION MAGAZINE *
95
IEEE Robotics & Automation Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2022
