Graphics Reference
In-Depth Information
Fig. 9.1
Picture of the Sig Rascal 110 ARF remote-controlled aircraft and a view into the payload
bay with the PC-104 board
the transmission can be improved without increasing the bandwidth, then payload
cost and communication distance do not have to be sacrificed. Aggressive image and
video compression already reduce the bandwidth dramatically, but depending on the
application, further reduction by several orders of magnitude is possible.
This chapter describes a UAV system that produced “actionable information”
through embedded computer vision where a traditional system would have been
limited by image quality or operating range. Figure
9.1
shows the UAV with its com-
mercial off-the-shelf (COTS) and custom payload (see Sect.
9.2.1
). The embedded
video analysis (see Sect.
9.3
) prioritized image regions of interest (ROI) for transmis-
sion based on content and thereby reduced the bandwidth requirements by a factor of
500 without impeding its operational “quality” (see Sect.
9.5.6
). This was possible
because searching for a specific object—vehicles—was the primary goal for flying
the UAV. This is a common need during desaster recovery, for search and rescue, and
in military operations: operators are searching for people, vehicles, or ships lost at
sea, for example. If likely image areas of such objects can be identified on the aircraft
or satellite, these areas of interest can be transmitted with higher priority and/or with
higher resolution (see Wilcox [
1
]) than the rest of the image.
It might even suffice to merely send the coordinates of a detection, a tiny frac-
tion of the size of an image. In tactical military operations, timeliness of “actionable”
information determines the usefulness of a UAV sensor and can determine the success
of a mission. Yet the above-stated “law” limits either availability (due to cost, size,
or logistics), range of operation, or image quality. Additionally, to benefit from tech-
nological improvements to the sensors (camera resolution and speed are increasing
rapidly), the downlink bandwidth would have to keep pace. Storing data on the UAV
is an option for some applications, but when timeliness or information sensitivity (in
case the UAV is lost) are issues, streaming is preferable over storing.
Most remote-controlled aircraft for hobby use differ in several ways. Most impor-
tantly, they do not send high-resolution imagery or video to the operator, only what's
required for navigation and control. That is, an operational restriction that the system
described here attempts to remove: images had to either be small, low resolution,
or be transmitted with growing latency or even downloaded after UAV recovery.
