IEEE Robotics & Automation Magazine - March 2022 - 99

TC SPOTLIGHT
Verification of Autonomous Systems
By Dejanira Araiza-Illan, Michael Fisher, Kevin Leahy, Joanna Isabelle Olszewska, and Signe Redfield
T
he robotics and autonomous
systems communities have
seen a significant and rapid
increase in both the development
of robots and vehicles for
commercial use and in using these
systems across a wide range of novel
applications. As these robots, vehicles,
software, and even embedded devices
move toward much greater autonomy,
techniques for verification, providing
much higher confidence than usual,
are becoming required. However, the
analysis and evaluation processes
used for traditional systems must be
significantly enhanced to provide in -
creased confidence in this next wave
of autonomous systems. The need
for well understood and effective
verification techniques will become
even vital, as we move to commercial
applications that rely on complex ar -
tificial intelligence technologies, and the
utilization of these systems in safetycritical
scenarios.
There are a growing number of
research developments concerning the
verification of complex systems that can
all impact upon this problem. These are
clearly of relevance for designing, constructing,
and deploying autonomous
systems, but also have importance to
psychology (e.g., social robotics), philosophy
(e.g., machine ethics), and law
(e.g., certification). Furthermore, constructing
autonomous systems without
strong behavioral guarantees can lead to
serious outcomes and may consequently
Digital Object Identifier 10.1109/MRA.2022.3143966
Date of current version: 22 March 2022
hold back the widespread adoption of
these systems. As the research is currently
fragmented and often not well
publicized, this technical committee
(TC) aims to coalesce this activity, drive
the research agenda forward, and instill
the necessity for verification firmly
within industry, government, standards,
and the public.
Background
Following a number of workshops and
collaborations, the Verification of
Autonomous Systems Technical Committee
(TC-VAS) was established within
the IEEE Robotics and Automation
Society in 2019. It has now developed
into a large community, with a mailing
list of over 490 unique individuals (493
unique people and 529 emails, as
checked on 23 October 2021), monthly
webinars, workshops, and input to standards
activities.
What Is " Verification " ?
The term verification encompasses a
range of techniques meant to assess,
with varying degrees of strength,
whether a particular system matches its
required behavior. Essentially, it consists
in providing evidence that some system,
S, matches its requirements, R. Since the
borders between verification and validation
can be blurred, this TC takes an
inclusive approach to such classifications.
The problem of defining requirements
that correctly and completely
describe the actual desired behavior is
likewise included.
Verification techniques are often categorized
into those that have a basis in
mathematical logic/proof and those that
rely on more empirical approaches.
Within each of these classes, there
remain many options. For example,
within the former (termed formal verification),
one might employ [8]:
●
Proof: where a formal proof is carried
out to establish that R, encoded as a
logical formula, follows from the logical
description of the behavior of S
● Model checking:
whe re
R is exhaustively
assessed
against a representation
of
all possible
e x e c ut i on
paths of S
● Runtime verification:
where
R is assessed
against the
system S as
it is executing
[10].
All of these
The term
verification
encompasses
a range of
techniques meant
to assess, with
varying degrees of
strength, whether
a particular
options also have
probabilistic versions,
many of
which have become popular in recent
years [2], [7]. There is also a range of
empirical verification approaches,
such as:
●
system matches its
required behavior.
Software testing: where R is checked
on a subset of the possible executions
of S
●
Simulation-based testing: as above
but where a simulation of the real
environment is used for environmental
modeling and interactions [1]
MARCH 2022 * IEEE ROBOTICS & AUTOMATION MAGAZINE *
99
IEEE Robotics & Automation Magazine - March 2022

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