Aerospace and Electronic Systems Magazine November 2016 Tutorial X - 102

Sense and Avoid for Unmanned Aircraft Systems

Fig. 16.

Architecture of multitarget tracking system (adapted from [9]).

operative tracks are often called "transponder tracks" and "skin
tracks," respectively. The need to track confirmed obstacles while
surveilling the entire sensors' FOV makes this a typical multitarget
tracking application [9, 33, 34], thus following a "classical" conceptual scheme as the one depicted in Fig. 16.
While this architecture can be applied to different SAA systems and operating scenarios, implementation choices, complexity, and challenges related to each block may change greatly. As
an example, association issues can be highly mitigated in cooperative architectures that can exploit Mode S or ADS-B information for aircraft identification, while the problem gets much
more challenging for ground-based sensing and relatively dense
environments.
Within this tutorial, we focus on some peculiar aspects that
characterize tracking systems for SAA applications, such as:
1. measurement formation
2. navigation system effects for onboard noncooperative SAA
3. aspects related to data association and state estimation
4. data fusion potential, architecture, and challenges (also considering ground/onboard and cooperative/noncooperative fusion).
In general, a peculiarity of tracking for SAA is the fact that
the interest is focused on the future position of the obstacle with
respect to the UA, much more than on its current position. Moreover, the main interest is on real and/or estimated near collision
conditions, i.e., on all the conditions that should or will activate
an avoidance decision. Thus, tracking performance in noncollision
conditions is of relatively limited interest, as long as it does not
lead to false conflict detections.
Measurement Formation: Measurement formation lies on the
boundary between sensing and detecting. An example of a typical
measurement formation task which may be challenging in SAA
scenarios is related to ground clutter filtering for radar-based onboard sensing systems.
Following [14], this task can be carried out in different ways:
1. using range rate and UAS motion information to isolate ground
echoes when the radar provides Doppler measurements.
2. estimating object velocity during a tentative tracking phase.
3. establishing an altitude threshold and removing all the returns
that apparently come from a lower altitude.
Concerning the latter strategy, the choice of the threshold has
to be related to the uncertainties in estimating apparent altitude of
radar echoes. Since these uncertainties depend on measured range,
102

a single scan criterion can be based on echo position in range vs.
altitude plane.
In practical cases, introducing the typical uncertainties of onboard
radars, when flying at low altitude it comes out that, at relatively large range, it is very hard to isolate low altitude aircraft and
ground objects with simple single scan strategies. Radar installation (e.g., in the nose or above the wing), antenna beamwidth and
steering angles (especially in elevation), and scan rate, play a key
role in this framework.
Navigation System Effects for Onboard Noncooperative
SAA: From the application point of view, the concept of system
accuracy is relevant to tracking outputs rather than to sensor-based
detections; in particular, quality of situation awareness information
has to be measured on the quantities that are directly used for threat
level estimation.
While a main factor influencing tracking performance is sensing accuracy, other effects have to be taken into account, such as
the valid measurement rate, tracking filter tuning and the capability to detect obstacle maneuvers, and the additional uncertainties
introduced by ownship navigation system.
The trade-off between accuracy for nonmaneuvering tracks
and early detection of obstacle maneuvers (thus reducing the risk
of track loss), and the benefits brought by more frequent measurements, are concepts that are commonly found in tracking implementations.
Regarding the navigation system from a practical point of view,
attitude estimation accuracy can have a major impact on SAA performance, both for single sensors and multisensor architectures.
Attitude measurement error noise can worsen raw detection accuracy, thus greatly reducing or even nullifying the advantage of
integrating high-resolution sensors. The detrimental effect of attitude measurement noise is much less significant in the case of
coarse obstacle detection sensors. Error biases on attitude angle
measurements (e.g., in estimation of magnetic heading) have little
impact on collision detection performance because, though generating angular position offsets, they do not affect relative velocity
estimates. The same conclusion holds for angular error biases due
to residual misalignment with respect to aircraft body axes, in the
case of a single sensor. However, instability of attitude error biases
can affect angular rate performance by generating erroneous rate
estimates, with consequent loss of situational awareness. As an example, a fast change in heading error bias would result, at tracking level, in an erroneous azimuth rate estimate. For multisensor
implementations, instability of error biases can also affect relative
alignment accuracy if a flight-based procedure is implemented.
If one takes technological trends into account, the recent development of relatively low-cost MEMS (micro-electro-mechanical
systems) inertial units with FOG-level (fiber-optic gyro) performance can play a key role for SAA onboard small platforms.
Data Association and State Estimation: Track-measurement
association issues are highly dependent on the considered scenario
in terms of airspace density, sensing resolution, and sensors location. As an example, noncooperative SAA for large UAS can be
assumed to be carried out in relatively low density airspace, thus
not needing very complex techniques for unambiguous association. On the other hand, possible implementation of SAA at low

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