Aerospace and Electronic Systems Magazine November 2016 Tutorial X - 90

Sense and Avoid for Unmanned Aircraft Systems
It is also interesting to verify how range requirements are relaxed when nonfrontal encounters are considered. To this aim, the
previous numerical modeling approach is used assuming that the
UAV is flying at 50 m/s, load factor during maneuver equal to 1.3
(bank angle of the order of 40°), and three possible obstacle velocities (65, 95, and 125 m/s). As before, τd is 5 s and the minimum
required distance is 500 ft.
The resulting detection range requirements are plotted in the
polar diagram shown in Fig. 12. In any condition, detection range
requirements are significantly relaxed considering nonfrontal collision geometry, basically due to the smaller range rate corresponding to given obstacle velocities.
It is worth underlining how detection range requirements are
linked to other parameters. In fact, these requirements change
within the FOV, while measurement rate and the trend of probability of detection as a function of range play a key role in determining the delay time.
As better clarified in Section IV, the probability of a confirmed
detection, i.e., the probability of a confirmed track corresponding
to an obstacle, is indeed the most important quantity from the application point of view. In practical cases, worst case assumptions
are often adopted to define sensing range requirements in given
scenarios.
Field of View: FOV can be defined as the angular sector surveilled by the sensing system. FOV requirements usually apply only to noncooperative architectures, as cooperative systems
are often omnidirectional and, thus, collect information from the
whole space surrounding a UA, at least within a given angular
range in the vertical plane.
Regarding FOV of airborne noncooperative sensors, a midterm
option similar to that of manned aircraft has to be selected. Elevation and azimuth angular spans can, thus, correspond to the recommended size of the cockpit window of a manned aircraft (30° in
elevation and 220° in azimuth). Smaller angular dimensions have
been proposed as a near-term compromise: as an example, 20° in
elevation and 180° in azimuth [14]. Though the concept of an angular FOV similar to the cockpit was initially derived from the
concept of an equivalent level of safety with respect to manned
aircraft, it is worth noting that a FOV that has the same shape as
that of manned aircraft does not imply the same level of safety,
since human pilots must keep their boresight in the direction of
the aircraft speed and since they do not scan the whole FOV with
regular frequency. In fact, the computations on detection range
have already shown that collision risk varies with intruder azimuth
angle, which is confirmed by the statistics that show that the large
majority of conflicts come from the central part of the FOV.
Sensing Accuracy: Intuitively, sensing uncertainties are related
to the performance of an SAA system in terms of probability of
missed detections and false alarms. In other words, sensing accuracy is closely related to the "quality" of the situational awareness
generated by the system. Indeed, information quality does not depend only on obstacle detection sensors/systems, but also on the
navigation system, the measurement rate, dynamic modeling aspects, etc. Thus, the relation between sensing accuracy and SAA
performance is better understood at the conflict detection level.
These aspects are discussed in Section IV.
90

Fig. 12.

Polar diagram of minimum sensing range variation as function of
obstacle azimuth angle for different obstacle velocities.

However, it is worth emphasizing that a fundamental difference exists between cooperative and noncooperative systems.
Noncooperative sensing systems produce obstacle information
expressed in terms of range, range rate, and/or angles. Thus, measurement accuracy is expressed as range and angles error. If one
considers cooperative systems such as ADS-B, based on the nature
of the Global Positioning System (GPS) solution, uncertainties
are expressed in terms of linear coordinates, both in position and
velocity. Additionally, traffic alert and collision avoidance system
(TCAS) (see Section III-B) surveillance has a hybrid nature due to
the fact that it includes both range and bearing errors, and altitude
errors which are actually linear errors.
This difference is of key importance when evaluating how
sensing uncertainties affect collision avoidance safety.
Also, there can be a logical link between sensing accuracy and
avoidance strategy, especially when the horizontal and the vertical
performance are very different from each other. TCAS resolution
is a good example for this concept.
Measurement Rates and Latencies: Measurement rates and latencies are closely related to the other requirements/specifications.
In fact, measurement rates and probability of detection are key
factors that determine the delay between initial detection and execution of avoidance maneuvers. Furthermore, measurement rates
have a strong influence on tracking performance, also because
larger measurement rates enable early maneuver detection without
needing ad hoc tuning for tracking filter parameters (which could
negatively impact tracking accuracy in nonmaneuvering phases).
These considerations are mainly applicable to noncooperative information sources.
Integrity: In general, integrity refers to the confidence in sensing information. The notion of integrity for cooperative systems
resembles the one adopted in navigation applications, i.e., integrity
typically represents the probability that a measurement falls beyond some limit characterizing its normal operation without having a timely warning from the sensing system. It can be extended
to noncooperative systems, where it can be associated to the confidence in the presence of a real intruder, to the statistical consistency of track estimates, and to the probability of track loss and of

IEEE A&E SYSTEMS MAGAZINE

NOVEMBER 2016, Part II of II

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine November 2016 Tutorial X