[squeak-dev] Autonomous Shark-Monitoring Drone

Eliot Miranda eliot.miranda at gmail.com
Wed Oct 7 15:54:55 UTC 2015


Dear Friends and Colleagues,

    as you may know, Sharks, as apex predators, are vital to maintaining
healthy marine ecosystems, and at the same time, their populations are
plummeting due to human actions.  It is estimated for example that the
population of pelagic oceanic white tip sharks is reducing by 17% per year
[1] and I've heard (can't find a reference) that populations in the eastern
indian/western pacific are at 1% of normal levels.  Such reductions in
populations create "trophic cascades" that produce wide-ranging changes in
populations of different species all the way down the food chain [2].  And
the marine ecosystem is a key source of human nutrition; it comprises
between 13% and 17% of global human protein intake [3].

  As you may also know, there is currently a shark attack crisis in New
South Wales [4].  While most people in the region oppose killing sharks in
response to the crisis, existing solutions, netting and culling reduce
those same threatened populations of sharks upon which the sustainability
of marine food supply d ecosystems depend [5], and are arguably ineffective
[6].  Apparently the most successful approach at avoiding attacks is the
use of human spotters, as used in Cape Town, where people in tall towers
scan the sea close to shore [6].

  But please watch this Youtube video
<https://www.youtube.com/watch?v=q2U3gjwJfS4> [7] from Pismo Beach,
California.  The shark is spotted at about 1:20 into the video.  This
drone, a phantom 3, is sending live video back to the operators, who are
using remote control.  What we can see from this video is that the point of
view of drones is far superior to that of spotters.

  My first thought is that autonomous drones could provide a cheap and
scalable solution to patrolling beaches to prevent shark attacks.  I expect
that processors like the Pi 2 have easily enough processing power to both
plan and execute search patterns along beaches, and perform the image
recognition necessary to reliably detect potentially dangerous sharks.  A
drone might also be able autonomously to visit surfers and swimmers near to
the shark and warn them, either by some signal such as flashing red LEDs or
an audible message (language issues notwithstanding).  The drone would have
to be able to identify swimmers and surfers in the water (not easy; sharks
confuse seals and surfers all the time), but computing an optimal route to
visit suspected swimmers should be relatively easy :-).

  I imagine that sooner or later it will be possible to construct cheap
rugged solar powered docking/charging shelters that drones could depart
from and return to, to charge and shelter from the elements after patrols.
Satellite communications could provide status reports for maintenance.

  My second thought is that as a community, we probably have all the
necessary expertise to construct a working prototype that at least
demonstrates feasibility.  Such a prototype would have to be able to fly
above the ocean along a beach under its own control for a useful period of
time, at least  15 minutes, and demonstrate that it can identify a shark in
the water (we could use a rubber shark
<https://www.youtube.com/watch?v=9dghbyBaQyI> for testing ;-) ).

Our community includes people doing image recognition with the Pi, radio
control experts, users of 3D printers, and some exceptional programmers.
If a small, strong group could be formed with the relevant expertise we may
be able to develop a prototype in a short enough time frame to be
relevant.  Of course, availability of time is a big issue. I couldn't
contribute more than a few hours a week.  But nothing ventured, nothing
gained.  So I'm writing this message to the Squeak and Pharo communities,
and bcc'ing a few friends I know that have relevant expertise to ask for
two things; first, for good information on scoping out the project,
possible technologies, power budgets, costs, what is available
off-the-shelf (both in hardware and known algorithms) and what we would
need to construct ourselves, and second, for volunteers.  Who among us are
really excited by this project, have relevant expertise and are motivated
to make a contribution?


[1] www.sharkadvocates.org/cites_4sharks_owt_fact_sheet.pdf
[2] http://www.globalshark.ca/pressmaterial/cascading/fig1_web.pdf
[3] http://www.who.int/nutrition/topics/3_foodconsumption/en/index5.html
[4] http://www.bbc.com/news/world-australia-34398516
[5]
http://www.theguardian.com/environment/2015/sep/29/shark-hit-australian-community-opposes-cull-research-finds
[6] http://www.nonswsharkcull.net/latestnews/tag/shark_spotting/
[7] https://www.youtube.com/watch?v=q2U3gjwJfS4



crazier ideas:

Infra red cameras are becoming cheap and easily available.  Species like
Great White and Bull Shark have elevated metabolisms, effectively they are
warm blooded, so these two facts may allow spotting at night.

Sharks are extremely sensitive to electrical fields; maybe some kind of
transmitter could be fitted to a drone, e.g. via a wire suspended in the
water, that could generate a field that could direct the shark away from
swimmers

_,,,^..^,,,_
best, Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20151007/d4fc08ea/attachment.htm


More information about the Squeak-dev mailing list