noted earlier, coordination activity is joint activity, in
the sense that what one agent does depends at least
partially on what others are doing. Joint human-machine activity is relevant to the theme in this article because “in sophisticated human-agent systems,
the underlying interdependence of joint activity is the
critical design feature” (Johnson et al. 2013).
Joint activity poses new requirements for sensing,
planning, communication, and other cognitive
capabilities that are based on coordination to
manage the interdependencies among human and
Metrics for Autonomy
To evaluate and compare designs for autonomy
within the conceptual design process, it is necessary to
identify a set of quantifiable performance metrics.
First, autonomy is a set of capabilities for exhibiting
goal-directed behavior. Therefore, we define an effectiveness
metric, namely, the degree to which the system exhibits
goal-directed rationality at a required level of responsiveness. By responsiveness we simply mean that it
does not take too long to accomplish its goals.
The second metric often associated with autonomous systems is robustness: the system is effective in a
wide range of operating conditions. This does not
necessarily require the system to exhibit goal-directed
rationality on its own in all operating conditions; for
example, under certain conditions, if it simply requests
assistance from a human operator, that decision is
sufficient to exhibit robustness.
The final metric is safety. Within the range of behaviors derived from mission requirements, the vehicle will not deliberately or accidentally harm itself,
humans, or its environment.
More metrics for autonomous system performance
can be proposed, but these three are sufficiently
comprehensive for us to confidently use them in the
Conceptual Design for Autonomy
In this section we combine the ideas of the previous two
sections to provide a high-level description of the role of
autonomy in conceptual design. More specifically, we
focus on a design from a scratch problem, rather that
what Raymer (2012) called a derivative design prob-
lem, one in which an existing detailed aircraft is
modified to be autonomous. As the purpose of this
article is to incorporate considerations of autonomy at
the conceptual design phase, derivative design, al-
though potentially important to study in the context
of autonomy, is not an issue to be discussed here.
First, the high-level purpose of the mission gives
rise to cognitive capability requirements for a vehicle.
For example, consider a mission in which an sUAV is
required to find small fires within an abandoned build-
ing. A core requirement for this application is that the
vehicle can navigate in a small indoor space. Another
requirement is that the vehicle can sense in a cluttered,
smoke-filled area with no GPS. Constraints induced by
Figure 1. Schematic of an Operational Architecture.
Command/data loops within a system that integrates human decision making with vehicle autonomy.