[Dprglist] Who needs GPS? Israel tests world's first drone that finds its destination without GPS

[Murray Altheim]

On 12/02/21 1:00 pm, Carl W Ott Jr via DPRGlist wrote:
> 
> *Israel tests world's first drone that finds its destination without GPS*
> https://me.mashable.com/tech/12869/israel-tests-worlds-first-drone-that-finds-its-destination-without-gps <https://flip.it/smEe-E>
   "...the drone entirely relies on its own judgment..."

   "Being empowered to identify its surroundings in real-time via machine
    learning, the delivery vehicle can function like a human pilot."

It pretty much sounds like this NaviSight software does exactly what a
BGM-109 Tomahawk Cruise Missile did in 1983. If I remember correctly,
one of the Tomahawk's first successful test flights was launched from
an Air Force base in the continental US (I think Vandenberg in northern
California), flew at low altitude all the way up to Alaska and right
into its target, a specific window in a building that had been loaded
on board as an image.

For navigation, the missile mapped its visual inputs to maps and fixed
images stored internally:

    TERCOM – Terrain Contour Matching. A digital representation of an area
    of terrain is mapped based on digital terrain elevation data or stereo
    imagery. This map is then inserted into a TLAM mission which is then
    loaded onto the missile. When the missile is in flight it compares the
    stored map data with radar altimeter data collected as the missile
    overflies the map. Based on comparison results the missile's inertial
    navigation system is updated and the missile corrects its course.
    TERCOM was based on, and was a significant improvement on, "Fingerprint,"
    a technology developed in 1964 for the SLAM.

    DSMAC – Digital Scene Matching Area Correlation. A digitized image of
    an area is mapped and then inserted into a TLAM mission. During the
    flight the missile will verify that the images that it has stored
    correlates with the image it sees below itself. Based on comparison
    results the missile's inertial navigation system is updated and the
    missile corrects its course.

No GPS. And no claims of "judgment" or "empowerment", no AI or Machine
Learning. Just very solid navigational engineering.

   "[artificial general intelligence] mostly seems stuck on the same
    issues in reasoning and common sense that AI has had problems
    with for at least 50 years."*
                                                    -- Rodney Brooks

Being a bit of a fan of Rodney Brooks, I can't help feel that the overuse
of "AI" and "Machine Learning" as marketing fluff must soon be reaching
its peak (like "peak-beard"), except this has been going on for what must
now be at least four or five decades, and has for all that time sounded
remarkably consistent, a bit like self-driving cars.

Machine Learning suggests there is learning done by a machine, but I
don't see that NaviSight is doing any learning. Marvin Minski would
call Machine Learning "suitcase words": we apply our own experience to
the term, ascribe our own beliefs, rather than consider what's actually
happening. It seems we've almost come to believe anytime a marketing
exec mentions "AI" that there's some kind of functioning "Artificial
Intelligence" without actually considering what that entails, i.e., a
functional definition of AI or whether something has reached whatever is
that generally-unknown threshold. Brooks and others in the field long for
the day when journalists and the public abandon the idea of the Turing
Test as a legitimate measure. It makes a good movie. Elon Musk is
apparently "terrified" of AI, which suggests despite all his money he is
not a very good thinker. And Jeff Bezos' greatest hope is that we'll all
soon live in outer space. I say to him: please go!

...not to put too much of a damper on flying delivery drones in Israel,
I actually had a conversation on the subject of AI/ML hype in my office
just yesterday...

Cheers,

Murray

----
* The Seven Deadly Sins of AI Predictions, by Rodney Brooks
     https://www.technologyreview.com/2017/10/06/241837/the-seven-deadly-sins-of-ai-predictions/
   Predictions Scorecard, 2021 January 01
     https://rodneybrooks.com/predictions-scorecard-2021-january-01/
...........................................................................
Murray Altheim <murray18 at altheim dot com>                       = =  ===
http://www.altheim.com/murray/                                     ===  ===
                                                                    = =  ===
     In the evening
     The rice leaves in the garden
     Rustle in the autumn wind
     That blows through my reed hut.
            -- Minamoto no Tsunenobu