Avionics News February 2022 - 32

UNIVERSAL AVIONICS' APERTURE
Continued from page 31
avionics and sensors on the aircraft. It virtually creates an
LPV or GPS approach to runways that don't have them. "
That's just one example of what it can do. Suffice it to
images from the aircraft's EFVS, other cameras, and sensors,
Aperture will do the same for pilots. "
Aperture: What a bright idea
Yahav noted that while Aperture is the first system of
its type developed for business and general aviation, the
concept is not new on the high-end military side.
" A number of industries heavily inspire the whole
artificial intelligence concept through their use of content
analysis and application technologies, " he said. " If you
look at the new fighters like the F-35, advanced weapons
systems, and some helicopters, they rely heavily on image
processing algorithms for navigation and target acquisition.
" In the F-35's system, the pilot can virtually see through
the airplane. Aperture will enable a similar capability by
digitally stitching together images from many different
sensors. "
While collecting video information from the aircraft's
cameras and knitting them into a coherent set of real-time
images is impressive in itself, Yahav said those capabilities
are just part of what Universal Avionics' Aperture will bring
to business and general aviation cockpits.
" The more-advanced functions are things like the ability
for the system to analyze information from those videos
and deduce information for the pilot, " he explained. " For
example, the camera looking out over the wing or tail and
alerting the pilot of a bird strike or the buildup of ice.
" Basically, the system compares 'normal' imagery to the
current video and determines if there is a problem. It's very
simple processing. But the system has many more advanced
capabilities it can provide to the flight crew. "
Yahav explained that among those " more-advanced
capabilities " are the ability to spot obstacles on the taxiway
or runway, as well as other aircraft movements in the air
and on the ground - including unmanned aircraft systems
and drones without transponders.
" Today, we are also showing that we can determine and
display the aircraft's position on the ground and in-flight
based on the collected video's data, " he added. " We call
it our Video Positioning System. It determines how the
aircraft is positioned oriented toward the active runway.
" On an approach, it can determine the aircraft's position
and give you feedback showing if you are above, below or
off to the side of the ideal approach path. At the same time,
it's collecting and coordinating information from the other
32 avionics news * february 2022
say, anytime the aircraft's various cameras and sensors can
" see " something, the system's high-end video processors
can combine that information and present it in a useful and
easily understood way to the pilot.
Machine see, machine do
Yahav wouldn't share any specific details of Universal
Avionics' skunkworks level information, but he did offer a
dumbed-down description of how Aperture works.
" As I said, we got a lot of ideas from the automotive
industry - in particular, the advancements with autonomous,
self-driving cars, " he said. " That's a lot more complex
environment than aviation is, by the way. For our
application, we developed our own proprietary algorithms to
handle the image processing via machine learning.
" If you want to teach a system every different situation
or scenario it will encounter using current methodology, it's
a monumental amount of work. We found a more-efficient
way to teach was to 'show' it examples. We used supervised
learning using images of things it will encounter - vehicles,
obstacles, other aircraft, drones, UASs, or whatever.
" Next, we developed a layer on top of that, which says
something like, 'If you see a drone 90 degrees off of the
left wing, what do you do? The whole idea of training the
system to work in an aviation environment has taken a lot of
technology and know-how. In the end though, we now have
a specific learning process and set of applications that are
dedicated to the commercial aviation market. "
Once you have collected and the learned machine
has analyzed said data, the next step is to implement its
augmented reality capabilities. That's when it puts it all
in play to present the " complete " situational picture to the
flight deck.
Yahav also said that to ensure Aperture will deliver
information of the highest quality, accuracy and reliability, the
entire system will be certified at Design Assurance Level A.
Speaking of systems, while the Aperture platform is
primarily a video collection and analytics tool, it needs to be
integrated within the installed camera and aircraft sensors to
ultimately deliver all of its capabilities.
As described by Yahav, a typical Aperture installation
will include an EVS camera, multiple airframe cameras, and
inputs from a host of avionics and sensors.
Continued on page 50
Avionics News February 2022

Table of Contents for the Digital Edition of Avionics News February 2022
