Aerospace and Electronic Systems Magazine November 2016 Tutorial X - 101

Fasano et al.
other purposes (such as marine or avian, i.e., bird-tracking radar),
though these may be reused and optimized for detecting low-flying
aircraft and obstacles.
Bistatic or multistatic radar works on the same principle as
monostatic radar but places the receiver and transmitter in different locations. This radar technology is one of the oldest sensing
technologies; the first radars from the 1930s were bistatic. These
radars detect objects by processing reflections either from a
known transmitter for an active method, or from noncooperative
sources of illumination in the environment, offering a passive solution. As with monostatic radar, improvements in advanced signal processing and computing power have significantly increased
capabilities. Unlike monostatic radar, bistatic and multistatic
radar is a maturing technology, that is still in the research and
testing phases.
The fundamental optical ground-based solution is the EO
camera, as a passive sensor. With the advances in technology,
very capable cameras have become so lightweight and low
powered that most UAs are equipped with one or more onboard
cameras, as standard capability. This capability can provide a
UA-centric view to the pilot on the ground. Cameras can also
be used as ground-based sensors, to surveille the environment
around the UA for target aircraft or obstacles. These cameras
have the advantage of being stationary (less motion-based noise
to deal with), and not being constrained in SWaP. The disadvantage of these cameras, like the airborne ones described in
the previous section, is that they do not sense distance-to-target
information directly. Distance information can be inferred from
the size of the target, through video processing, or relying on
another sensor to provide range information. As a result, most
camera-based solutions are a fusion with other sensors, or require substantial processing (which has improved significantly
over the past several years).
Acoustic detection utilizes passive "listening" techniques to
acquire the presence and location of other aircraft. Acoustic listening devices have been used for aircraft detection since World
War I, with a proven history and use in detecting low-flying
manned aircraft to a range of 10 km and greater. By recognizing the specific acoustic signatures of aircraft, these sensors are
capable of classifying, localizing, and tracking specific aircraft
types, and can filter out unwanted noise. This is a mature technology, though it still may require optimization and validation
to prove safe in the range of UAS operational environments. It
is very small and lightweight, so it may be used onboard the
UA, as well as in a ground-based central or distributed configuration.
Two technologies are active acoustic (principally fulfilled by
SONAR) and active light (principally fulfilled by LIDAR).
Both of these technologies have range limitations that limit
their effectiveness to hundreds or thousands of feet, respectively. If the UAS mission falls within that range, then these
would be potentially useful components of an SAA solution.
However, this range falls within the definition of "visual line
of sight" (VLOS) from the pilot to the UA, and many countries
are establishing rules to permit small UAS to fly within VLOS
without an SAA solution. To be able to fly beyond VLOS, senNOVEMBER 2016, Part II of II

sors and SAA will be required, and these sensing methods will
be less useful.

IV. DETECTING: SITUATION AWARENESS AND CONFLICT
DETECTION
Increasing confidence in the presence of real obstacles in the environment, and understanding if these obstacles can pose a threat to
UAS flight, is the scope of the detect task. This is usually carried
out by exploiting a target tracking and a conflict detection algorithm, which in some implementations can be fused together. Conflict is usually defined as an event in which the distance between
aircraft breaks the minimal defined separation criterion. These algorithms will be focused in this section.
Indeed, basic detecting strategies may foresee no threat estimation, meaning that the pilot or the flight control system is alerted
whenever a sensed target is entering a specific geographic area.
This approach is clearly suboptimal and it may generate a large
number of unneeded avoidance maneuvers. This section will mainly concentrate on more refined processing and decision-making
strategies.

A. TARGET TRACKING
Sensor-based object detection, i.e., the recognition of a cluster of
data included in the output stream of the sensor that can be associated to an object that is significantly different from background,
usually does not suffice for declaring a collision threat, and thus
does not represent a reliable source of situation awareness information. In general, this is due to several reasons, such as the following.
1. Several, or even most sensor detections can be due to clutter
and false alarms, instead of obstacles to be avoided. These returns have to be filtered out, while consistent sensor detections
have to be associated to generate confirmed tracks.
2. In most cases sensing data do not allow proper collision risk
evaluation, and thus have to be complemented with indirect estimates generated by a tracking algorithm. Estimates of obstacle kinematics represent a typical example in this framework.
Besides obstacle position and velocity, information about estimation uncertainty (usually provided by tracking algorithms
within track covariance matrix [9]) can be effectively used both
for conflict detection and avoidance purposes.
3. Sensors usually provide their output with unstable data rate
and missing measurements, while the user needs estimates at a
stable data rate and, in some cases, at a higher frequency with
respect to the raw data one.
4. Reducing sensor noise while adding robustness to missed detections clearly brings benefits for the SAA application.
Both for cooperative and noncooperative sources, and both
in the case of onboard and ground-based architectures, an SAA
system thus comprises a tracking algorithm as a core component.
Within architectures that include radars, cooperative and nonco-

IEEE A&E SYSTEMS MAGAZINE

101

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