IEEE Systems, Man and Cybernetics Magazine - October 2019 - 24

The Need for Safety and Trust
During the past few decades, significant progress has
been made in the development and deployment of autonomous vehicles, leading to the establishment of many
emerging industries. Sophisticated methodologies for
sensing and perception, innovative techniques for mapping gathered information, and efficient algorithms for
optimal decision making have been developed [1]-[3].
From the cruise control in our cars to the autopilot function in aircraft and the use of increasingly automated
robots in the manufacturing industry, autonomy is becoming a huge part of our daily lives. Autonomy is about performing certain operations without the involvement of a
human in the loop [4].
Testing procedures for fully autonomous cars and
trucks have been rapidly progressing, enhancing the promise and the commercial availability of autonomous road
vehicles in the near future. Likewise, other unmanned
ground vehicle technologies [5] have been under development with investments from civil, space, and military programs, which have led to advances such as the latest Mars
rovers and automated ground vehicles for factory automation. Furthermore, the aviation industry has a track record
of developing comprehensive technologies, sensors, and
instruments that have been integrated into different autonomous unmanned aerial vehicles (UAVs) [6]. These efforts
and technological innovations have been reinforcing the
development of new projects, such as autonomous road
vehicles, often referred to as driverless cars; air taxis for
on-demand mobility; and autonomous spacecraft for
future space travels.
An autonomous vehicle needs to perceive the environment to accomplish the assigned missions. For example,
an autonomous car intended for road travel must be able
to detect lanes, other vehicles in its environment, and
obstacles, such as pedestrians and traffic signs, while following the local traffic laws. UAVs need to adhere to flight
plans and deal with obstacles, such as tree branches and
transmission cables, as well as other aerial vehicles operating in the general vicinity. To achieve these expectations, autonomous vehicles should be able to navigate

TECHLAV

Modeling
and
Analysis

Reliable
Control and
Communication

T&E

Figure 1. The main thrusts of the TECHLAV Center

regarding the large-scale autonomous systems
of vehicles.

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE O ctober 2019

in an unstructured environment while detecting and
avoiding obstacles and planning and tracking an optimal
path toward their destinations [5]-[8]. These practices
should be followed by solid plans for the testing and evaluation (T&E) of the techniques to reduce possible failures
and accidents.
The World Health Organization has reported an average of 1.25 million deaths due to road accidents and
many times more injured individuals per year worldwide [9]. Air transportation has the potential to have
similar scale threats as aviation capacity increases
with more commercial aircraft operations, including
UAVs and piloted aircraft. The main concern relative to
the widespread-use of autonomous vehicles is the safety of people and the security of autonomous unmanned
systems (AUSs) and associated infrastructures. The
first death attributed to an autonomous car was reported in March 2018 [10] when a self-driving car in its fully
autonomous mode struck and killed a pedestrian on a
crosswalk. This suspended operation of self-driving
cars by the transportation company, Uber, while the
incident was investigated.
Operationalizing autonomy (OA) focuses on the largescale implementation of autonomous systems [11] and
their applications and commercial availability. To make a
transition from laboratory-scale technologies to operationalized service platforms, we need to build trust in AUSs in
terms of enabling machines to reliably conduct operations
with little or no human involvement. Currently, testing
autonomous systems is performed in controlled environments, generally following a series of prescribed protocols
[2]. To deploy these vehicles, developers and stakeholders
need to have high confidence in their safety, functionality,
and reliability. This requires the placement of reasonable
threshold levels regarding their successful performance,
with the improvement or creation of policies and regulations for their ethical use. It has been established that OA
is an evolving process that will improve autonomous capabilities over time. There are associated risks involved with
the deployment of AUSs; however, these risks are expected
to be reduced through improvements in the technology
and related policies.
This article offers a preliminary definition of OA and its
relevance in AUSs, especially in future defense and transportation infrastructures. A technical meeting at the
TECHLAV Center inspired the definitions and examples
provided here. The discussion was aimed at identifying the
challenges for OA and related technologies. The TECHLAV
Center was initially funded by the U.S. Department of
Defense (DoD) and the U.S. Air Force Research Laboratory as a Center of Excellence in Autonomy, in 2015 [12].
TECHLAV's research program is designed as an "umbrella
program." As depicted in Figure 1, the TECHLAV Center is
built upon three research thrusts for large-scale autonomous systems of vehicles: modeling and analysis, resilient
control and communication, and T&E.

IEEE Systems, Man and Cybernetics Magazine - October 2019

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - October 2019