January 28, 2021, 03:50:33 pm

AI Cessnas

Started by exarmyguard, August 28, 2020, 12:32:21 am

0 Members and 1 Guest are viewing this topic.

exarmyguard

This week, I read that a Cessna Caravan and Cessna 172 have been converted into autonomous aircraft. I believe the 172 was successful in its tests. I will try to locate the news articles and post them.

I was wondering the opinions of the members here of this new technology, it's possible effect on CAP and the future for pilots in general. Thank you.

Spam

This was actually done a decade or two ago, if I recall correctly. Conversion of manned aircraft to UAS config is not a new thing; the first tactical UAV was actually in WW1.  Years later in WW2, UAVs were built out of B-17s and other aircraft for remotely guided attacks, and the first TV-guided bombs were used in combat (I think, first, on an attack on the Italian fleet at Taranto, but I could check that).

So what does that mean for us?  One of the original strengths of CAP was its ability to leverage the existing tickets and skill sets of its volunteer pilots to serve DoD (actually Dept of War, then). The further we depart from that, requiring high tech which isnt mainstream, or which is "niche", we lose that leverage. Really, in the end, what capability do we lack that requires a new toy?  (Not - "hey lets buy the toy and then see what we could use it for").

R/s
Spam

Spam

August 28, 2020, 12:57:34 am #2 Last Edit: August 28, 2020, 01:25:23 am by Spam
Yeah, found them. See:

WW1
https://en.wikipedia.org/wiki/Siemens_torpedo_glider
https://en.wikipedia.org/wiki/Kettering_Bug


WW2
Allied use of UAV bombers
https://www.airforcemag.com/article/1110bombers/
Axis use of guided missiles, in a 1943 attack on Allied ships off Sicily:
https://en.wikipedia.org/wiki/Fritz_X
*Sorry, the famous RN attack on Taranto was with (manned) Fairey Swordfish aircraft at a murderous cost... I was confusing the two Med ops.  https://www.nationalww2museum.org/war/articles/taranto-november-1940

V/r
Spam

Eclipse


Barring significant societal disruption, the days of the human pilot are numbered.
What that "number" is, is hard to say, because there are variables at play that
will attempt to deny or slow the progression for sentimental and perceived risk
reasons vs. capabilities.

It would not surprise me if if was in 10 years on a wide scale, it would surprise me
if it wasn't within 20.  Unlike transporters, the tech already exists, it's just a matter
of refinement and acceptance.



exarmyguard

https://fortune.com/2020/08/26/space-x-tesla-reliable-robotics-autonomous-airplanes/

Will future cadets be groomed to operate UAS vehicles? Will CAP gear itself to the keyboard warriors of tomorrow?

Spam

The work I am engaged in right now (actually RIGHT now, with a briefing tomorrow on the CONOPS update) is to incorporate automation which would free human pilots to pilot by pointing, while automation policy managers provide coupled envelope protection, terrain and traffic avoidance, etc. Essentially that is how the Raptor FLCS works, by accepting a pilot "go here" command and by then managing surfaces, thrust, and systems to put the aircraft where he asks - rather than by moving control surfaces where he THINKS they should go.

The idea is to free pilots up to think tactically without worrying about hitting terrain/other planes, stalling, etc.

But some slower to adapt people are still thinking about flying NOE, manually, with all the inabilities of the human perceptual and musculoskeletal systems to react in a timely enough fashion to provide flight critical inputs. They are bucking the idea that automation should keep the stick, obeying all commands from a human except those which cause loss of the aircraft...

V/r
Spam

etodd

Its all good. A pilot can be sitting on the ground clicking a touch screen to fly cargo somewhere for his job.  "Its just a job."

And then on the weekends get in his Piper Cub to fly low and slow over the countryside with the wind in his face, stopping for a burger somewhere, and enjoy a beautiful sunset on his way back.

Like the truck driver, on the highway all week, that then enjoys getting on some back country roads on his Harley for fun.

Yes ... I'm seeing a new generation coming along whose idea of fun and adventure will be virtual on screen. Sad.
"Don't try to explain it, just bow your head
Breathe in, breathe out, move on ..."

PHall

You know Eclipse, you've been saying the days of the human pilot are over for years now,
but you have never, ever cited anything as your source for this.
So, how exactly did you come by this opinion?

Eclipse

Quote from: PHall on August 28, 2020, 03:46:46 amYou know Eclipse, you've been saying the days of the human pilot are over for years now,
but you have never, ever cited anything as your source for this.
So, how exactly did you come by this opinion?

You're kidding, right?




Paul Creed III

Quote from: exarmyguard on August 28, 2020, 01:29:46 amhttps://fortune.com/2020/08/26/space-x-tesla-reliable-robotics-autonomous-airplanes/

Will future cadets be groomed to operate UAS vehicles? Will CAP gear itself to the keyboard warriors of tomorrow?

Cadets are already being trained with the STEM kits and cadets can join CAP's sUAS program - I'm in a sUAS course now with a bunch of cadets.
Lt Col Paul Creed III, CAP
National Headquarters Cyber Curriculum Specialist
National Headquarters Photography Working Group
Instructor, Volunteer University
Akron-Canton Senior Flying Squadron Commander

TheSkyHornet

Quote from: Paul Creed III on August 28, 2020, 12:05:07 pm
Quote from: exarmyguard on August 28, 2020, 01:29:46 amhttps://fortune.com/2020/08/26/space-x-tesla-reliable-robotics-autonomous-airplanes/

Will future cadets be groomed to operate UAS vehicles? Will CAP gear itself to the keyboard warriors of tomorrow?

Cadets are already being trained with the STEM kits and cadets can join CAP's sUAS program - I'm in a sUAS course now with a bunch of cadets.

I really wanted to be in that course. I realized it was a month long, not a week. I just can't commit the time to it.

But I'm hearing that it's really educational. Like any course, dry at times, but overall very interesting material and really eye-opening for the people sitting in. There's a lot more to sUAS than flying toys out of boxes from a mission standpoint.

I don't know how long before we go even a quarter of Part 121 flights being AI. I think we're quite far away from that. I can see maybe the commercial 91 flights being a starter, perhaps air taxis even. That's all hypothetical rhetorical, though.

I think we do need to make sure that we're on the same page here when we refer to AI vs. UAS/Vs in this thread. There is a significant difference: remote-piloted versus fully autonomous (using an observer/programmer, not a pilot).

Capt Thompson

If we did eventually go to autonomous aircraft, I would think an observer/programmer wouldn't be enough to satisfy the FAA? I would think that person would need the ability to take manual control and pilot the vehicle should something go wrong, so in that case there would still be just as big a need for pilots no?
Capt Matt Thompson
Deputy Commander for Cadets, Historian, Public Affairs Officer

Mitchell - 31 OCT 98 (#44670) Earhart - 1 OCT 00 (#11401)

Spam

"Though Xwing says its system can handle a flight from takeoff to landing, humans are still needed for the planes to operate, particularly to communicate over radio with air traffic control officials".

Automation is half of a blended human/system continuum of capabilities along that spectrum. On the Folds automation spectrum from Direct Performer (level 1 - pilot full manual control of every task) to Executive Controller (level 4 - human monitor and intervene by exception) most of our CAP sUAS systems are around a 2, with automation for flyback to a point, etc. I don't see the Xwing system as described being much more given the limited data in their press brief (thanks for uploading the link, Guardsman, btw). So, when we use the term "autonomous" it is helpful to think in terms of automated capability sets vs manual capabilities reserved for humans.

One of the better references I've used lately (gonna help Eclipse out here with the citation stuff) is (U) Autonomous Technologies for Unmanned Aerial Systems (ATUAS) JCTD Operational Utility Assessment (OUA) report, available through SURVIAC, 96 TG/OL-AC, Wright Pat AFB. Dated Feb 2014. While distro D, it is very useful in terms of characterizing the significant problems left to solve in developing and fielding a safe system. Based on a series of exhaustive flight tests it gets down into very detailed recommendations (e.g. the cumbersome "inability to change the mission plan with an active spatial conflict" within tested merge timelines is both safety and suitability barrier for one system). So, the problems identified validate many of the real world problems identified to date with driverless cars and delivery UAVs, etc.

With something on the order of five fatalities and a number of non fatal automated vehicle mishaps already, the driverless car experience is to my mind a giant and somewhat fatal beta test of a still to be perfected system. As a guy with 30 years working in aerospace I don't consider the fielding of driverless cars to be a good model for aviation to follow. I would hate to be considered part of that Darwinistic process myself, and while I've accepted personal risk (aircraft carrier ops, night ops, etc.) in aircraft test, my family has NOT. Noting the pilotless Caravan in that video on approach over a muni airport and streets... if my family were killed by someone else's immature, prematurely fielded driverless tech, I would leave the funeral and head after the CEOs, engineers and regulators that were liable for aiming and releasing that immature tech at my family.

So, TL/DR - combining this ORM factor with cost and training reasons I don't ever see CAP using this for larger UASs or optionally manned aircraft, ever.

R/s
Spam

Eclipse

Quote from: Capt Thompson on August 28, 2020, 02:51:13 pmIf we did eventually go to autonomous aircraft, I would think an observer/programmer wouldn't be enough to satisfy the FAA? I would think that person would need the ability to take manual control and pilot the vehicle should something go wrong, so in that case there would still be just as big a need for pilots no?

CAP isn't going to get autonomous aircraft, that's the point.

When the tipping point is reached, the aircraft go away. By then they aren't needed as
much cheaper, smaller, and mission focused devices, likely lots more rotor and lots less fixed-wing
will be the norm.

O-Rides?  For what?  There's little to no need for traditional "pilots", so
no need to "feed the love of General Aviation", since GA ain't no mo.

ES?  Point and shoot UAVs will be doing that.

DR? See above.

No one with sense would pay to have the CAP fleet retro-fitted or replaced.  At best it's
aircraft without seats being run from an ipad at the same kitchen table as the
cell phone forensics guys. Most of the fleet is scrapped or sold as museum pieces
for recreational use in the same way as some people still own 57 Chevys.



Eclipse

August 28, 2020, 06:16:33 pm #15 Last Edit: August 28, 2020, 06:23:22 pm by Eclipse
Quote from: Spam on August 28, 2020, 03:28:06 pmWith something on the order of five fatalities and a number of non fatal automated vehicle mishaps already, the driverless car experience is to my mind a giant and somewhat fatal beta test of a still to be perfected system. As a guy with 30 years working in aerospace I don't consider the fielding of driverless cars to be a good model for aviation to follow. I would hate to be considered part of that Darwinistic process myself, and while I've accepted personal risk (aircraft carrier ops, night ops, etc.) in aircraft test, my family has NOT. Noting the pilotless Caravan in that video on approach over a muni airport and streets... if my family were killed by someone else's immature, prematurely fielded driverless tech, I would leave the funeral and head after the CEOs, engineers and regulators that were liable for aiming and releasing that immature tech at my family.

While this is a fair and understandable attitude, it comes from the perspective that those crashes were explicitly attributed to being driverless, which may or may not be the case. Someone running out in front of a vehicle unexpectedly, for example, has nothing to do with whether it was driverless.
In fact most developers of autonomous tech say regularly that the biggest risk to their vehicles is legacy vehicles driven by people, who are unpredictable and often crazy.

We live in a world where people block charging stations with combustion engince vehicles for...
reasons?

While at the same time the intertubes tell me...

There are Over 16,000 automobile crashes per day.

90 resulting in fatalities.

Per day.

Anyone who has done any cursory study, or even Google Research knows that
a significant number (probably the majority) of automobile crashes, and a non-trivial
number of aircraft mishaps, are caused by human error, caused often by distraction,
fatigue, or lack of ability in a type (car or plane).

In fact I'd hazard most GA mishaps are human error of one kind or another.

Autonomous operation reduces or eliminates many of the factors endemic to both ground
and air vehicles.

What if the number could be reduced to 5, but one of those is your sister in a driverless car?
90 vs. 5.  The lawyers have a field day and the actuaries have an easy choice.

The trolley problem is going to be more of an impediment then development.

There has never been a significant technology that didn't cost lives and jobs during
development and transition. That doesn't ease the pain of losing someone due to
a software glitch or system failure, but it doesn't change the argument, or the ORM
at the macro level, which is where the actuaries live.