Now that the Garmin Autonomi has been developed and certified the question of how much flying an autopilot can do has been answered. Everything. The Autonomi can autonomously select an appropriate airport, fly the approach, and land the airplane without human intervention or input.
The Autonomi system is intended for emergency use only, but it is the premier example of how capable automatic flight guidance systems can be. Even without autonomous landing capability, the modern autopilot can fly with incredible precision in all phases of flight. And over all but brief periods, the capable autopilot system will be more precise and smoother than the human pilot.
Garmin’s autoland technology is just the most visible proof that modern autopilots can do it all.
However, without redundant fail-operative systems, the human pilot must hand fly at least some of the time. How does the human pilot retain and practice the skills necessary for precision hand flying while still making best use of the autopilot system? That’s the question.
In single-pilot high performance airplane flying, the standard procedures and training are for the autopilot to fly nearly all of every flight. It’s required that the autopilot system be fully functioning on every single-pilot flight in jets, and on charter flights flown by a lone pilot. And you won’t get past a checkride at any major training facility as a single pilot without using the autopilot in all phases of flight and fully understanding all of its modes and capabilities.
There is historical precedence for autopilot use by a single pilot. When I first started flying jets 40 years ago, the standard procedure was for the pilot flying—the one actually manipulating the controls—to do nothing but scan the instruments and move the controls. If the pilot flying (PF) wanted to look at a chart, tune a radio, review a checklist, or do anything else except look at the flight instruments he handed the airplane to the other pilot.
It was a small step from the concept of the PF doing nothing but holding heading and altitude, the proper airspeed and course to having an autopilot do the same thing when there was no qualified human pilot in the other seat.
It makes sense to me that pilots flying IFR in piston airplanes would benefit from applying the same procedures. By using the autopilot to its fullest capability, the human is freed from the need to concentrate essentially 100% of his attention on maintaining the desired flight path.
Of course, the human must continuously monitor the autopilot actions and performance. But that’s much easier than hand flying full time. You have experienced this anytime you’re flying in the right seat—or sitting in the passenger seat of a car. It’s so much easier to see “what the other guy is doing wrong” from that position than when you’re the one with hands on the controls.
It’s also accepted practice that the autopilot fly approaches with weather near minimums. In fact, reduced minimum category approaches are based on use of automatic flight guidance. And autopilot flight is also a part of approval to cruise in the reduced vertical separation minimum (RVSM) airspace above Flight Level 290. Regulators recognized that when you’re descending to within 100 feet of the runway (or less on an ILS) or maintaining cruise altitude in the thin air above FL 290, only automatic systems have the consistent precision of performance.
But none of this addresses the fundamental question of how we human pilots keep our flying skills—particularly our instrument scan—polished while the autopilot does nearly all of the flying. One answer for pilots of jets is the FAR 61.58 requirement to be trained and checked annually. During that recurrent training you will be expected to use the autopilot fully, but you will also experience autopilot failures in the simulator, typically during critical, high workload phases of flight. If you demonstrate proficiency during those events your skills are up to date.
Jet pilots have to regularly check their scan in simulators, but what about light airplane pilots?
For pilots flying airplanes that don’t require a type rating, there is no FAR 61.58 to check our skills. The flight review every two years is only a review, not a check. And the review can be flown in any class of aircraft you’re rated for, not the most demanding. And it need not be flown under IFR standards even if you’re IFR rated. The flight review is better than nothing, but will only be as beneficial and revealing as the pilot and CFI want to make it.
Many pilots, including me, opt to hand fly much of the departure and climb to cruise. This is a higher workload phase of flight, often with configuration changes, intermediate level-offs, vectors, course changes, and speed adjustments. It’s a pretty good workout of your instrument scan and control precision.
But when the air is choppy and the ATC instructions are changing fast, I look at the instruments and wonder if I’m giving the passengers the best possible ride by hand flying. No matter how hard we concentrate and try to be smooth on the controls, none of us can match the performance of a modern, sophisticated automatic flight guidance system.
One reason is that the autopilot never scans. Dedicated channels continuously track each flight parameter, something our best scan simply can’t do. The word “scan” gives it away. The autopilot never scans because it is 100% dedicated to the task.
Modern autopilots also have very sensitive anticipatory capability. For example, when approaching an altitude capture the system obviously monitors altitude and vertical speed as we humans would. But it also has accelerometers and uses that information to calculate an intercept and level-off limited to no more than a few hundredths of a G, while never over- or under-shooting target altitude. Maybe I get lucky and match the autopilot smoothness sometimes, but it does it every time.
We’re watching these same questions and concerns play out on a very big stage with introduction of autonomous control of automobiles. How to keep the driver involved and in command while also reaping the benefits of automatic control are the burning questions. Sound familiar?
My hope is that with millions of participants in the move toward autonomous driving vehicles instead of the few hundred thousand of us pilots, we will learn more about human interaction and control of automatic guidance systems and all benefit.
Until that happens we’re each as pilots left with the question of how much hand flying practice do we need to be safe and competent while still fully using the safety and comfort capability of the automatic system. After all of these years of flying, I’m still not sure that I know for myself, much less deliver an answer for other pilots.
- Cirrus and Bonanzas dominated their eras - May 22, 2023
- Lead-free avgas STC—a historical preview? - November 21, 2022
- What Nexrad can—and can’t—tell pilots - March 14, 2022
Seems about time to start tracking the difference. PIC when flying, Manager in Command (MIC) when managing the autopilot.
Maybe “MIC” should be “RIP”. If the autopilot is more error free than the protoplasm pilot, and “redundancy” is needed… why have a MIC. Why not an AI based backup autopilot, or perhaps two? It’s a heck of a lot cheaper to have that double or triple redundancy than either get pilot training & maintain current qualifications or have a human pilot up front in the aircraft who gets a paycheck, fringe bennies, and retirement. And the AI (that’s artificial intelligence) is much less likely to bust the rules, get distracted, or go to sleep on those long and boring stretches in every flight.
everyone who thinks AI is the next big thing has either not done software development outside the lab or is selling a bridge in Brooklyn…AI is the magic that keeps you from having to develop use cases for every single eventuality…real world is a bit messier than most SW developers acknowledge…maybe if they got out a bit more often they’d understand that…how’s that autonomous driving going with pedestrians, construction zones or heaven forbid, poorly marked or snow covered roads?
Exactly. Seems about as “just around the corner” as the Moller flying car…
…and if it is ready for prime time, let’s try it out as a virus and spam blocker…
When AI gets good enough to dynamically defend against the millions of HI [human intelligence] cybercriminals, then I might let it drive my car. Meanwhile, I’ll keep my left hand on the wheel [yoke] and my right is available for the other tasks [throttle]. Both on the asphalt and in the air, the PIC is responsible for safe completion of the mission.
This goes hand in hand with the notion that PPL and IFR rating syllabus be split into “steam” and “glass”. There are pilots, myself included that have no desire and will never fly, a now 40yr old steam gauge airplane. There needs to be a rating (then emphasis and mastery) of one or the other. To continue to labor the point of a T-pack/wagon wheel scan to a young pilot that will only ever fly G1000nxi phase II and beyond just continues to date the FAA training into irrelevance.
I would personally gladly accept the “limitation” of not flying museum pieces in return for updated emphasis on the systems one is actually flying day to day.
Hi Steve,
There are precedents for your idea in transport category flying. For example, a pilot who doesn’t train and practice circling approaches during required recurrent training of FAR 91.68 gets a limitation on his certificate that he is not authorized to circle.
It would not be difficult to restrict pilots who only train and fly with modern glass cockpits to glass only, no steam gauge allowed.
Mac Mc
In my experience as a DPE, I see many applicants flying a G-1000 airplane that either hand fly or use the automation well. Their training only emphasized one or the other. That is a huge opportunity missed. I also fly the Airbus A320 family and both skills (and the lesser important flight simulator skills) are necessary for success. While I agree that one can practice by mixing it up, I would argue it should start on day one of training if so equipped.
Every 90 days, fly a entire sortie with both the autopilot and the flight director turned off.
I’m as student pilot, and my thinking, while pursuing my PPL is to work with steam gauges, and hand fly, 100% through achieving my PPL, then begin learning, with a CFI, glass, auto-pilot, and IFR, to fly current piston aircraft. I’m not interested in flying antiques at this time. My typical mission would be day-trips or weekends with my wife/family/friends to interesting destinations in my region of the country. Do you have any comments or suggestions regarding my intended path to becoming the best pilot I can be?
Perhaps my comment doesn’t quite fit into this discussion amongst modern-day aviators, because I’m an old steam-gauge has-been, use-to-be pilot. Nevertheless, I’ll toss in my two-cents.
Long, long ago, I had a PA28-161 with wing leveler, which I coupled to a LORAN that I installed. That was great! Best thing since sliced bread. However, after several months, I realized that with my over-reliance on the automation, I was losing my ability to navigate using VOR triangulation… and could find myself really uncomfortable, and even in dire straits, should any element of the automation fail. So, I forced myself to frequently leave the LORAN turned off, and rely on the VORs and charts to retain my pilotage abilities.
My advice to my grand daughter when she began flying (she now flys glass cockpit jets) was to first concentrate on learning to fly the plane, then learn to “fly” the computers… but to always work to better her ability in both realms.
Now, seems to me to add a “glass/steam” designations or endorsements complicates things, unless it’s associated with type ratings. But, by the time our current FAA could manage a decision, there won’t be enough steam gauges outside of those in museums left to matter.
Fini
Horses for Courses!
I’m a 50+ year Air Force (originally) trained pilot with the privilege of owning a 1980 T-210 for 30 of those years which we’ve now converted to glass and just added a Genesys 3100 digital a/p. I’m learning to fly the new way you describe Mac.
We’ve also got a 2004 (very VFR) Super Decathlon at a second home on a 2300’ turf strip. Hoping to keep my well honed steam gage skills while at the same time climbing on a new and fantastically improved horse while looking for an answer to your question! Thanks-seems like I’ve read your columns for almost as long as I’ve been flying-always thought provoking!
Thanks for the kind words, Don. Your new Decathlon sounds like loads of fun. I would think an airspeed indicator to keep tabs on takeoffs and landings on your strip and a slip-skid ball to stay in trim would be all the instruments you really need.
Bests,
Mac Mc
I don’t believe it is an all or nothing situation, either the human pilot flies or the machine pilot flies. As background, with any technological advance, it will keep advancing! It is hard to imagine, let alone predict with any accuracy, where the next level of that advancement is headed. It is worth understanding where the technology is today and how it differs from yesterday, to fully appreciate how it can be used. As a very brief example, technology called “autoland” has been around – large commercial aircraft have had the capability for some time to automatically land at a destination airport. That autoland requires the PIC to decide, input and store where that destination was and the aircraft route to get there. The PIC still has to watch out for adverse weather, select and communicate with the ATC facility which was appropriate for the aircraft location and generally keep an eye on things, including the autoland system.
The new generation of autoland like the above described “Autonomi can autonomously select an appropriate airport, fly the approach, and land the airplane without human intervention or input.” Not only that, it checks the weather, selects the airport and runway, chooses a course and altitudes to avoid terrain and bad weather, ensures the chosen airport runway is suitable for the aircraft and ensures there is adequate fuel to reach and land on that runway via that course. Further, while en-route Autonmi chooses the ATC frequencies which are appropriate for the aircraft altitude and location and audibly communicates intentions to ATC while at the same time audibly informing the PIC and passengers of flight status. Unlike the old generation autoland the new generation autoland is making decisions without human assistance, about where to go, how to get there and who to communicate with while on the way. But, currently it is either on or off, if needed it autonomously takes over control of the aircraft to provide emergency autoland.
I believe the next technology advance will be to in effect, keep an eye on the PIC even when there is no emergency. This technology now has the ability to be a constant flight instructor or co-pilot, quietly watching over the PIC to make sure everything the PIC is doing is within safe operation envelopes and appropriate for the aircraft, airspace, airport, communications and ATC requirements. Like a flight instructor, this new technology can be used to advise, caution, assist the PIC, or if desired and enabled, take over one or more of the flight tasks from the PIC. Assuming those choices are made available, the PIC will be able to practice while letting the autonomous machine flight instructor / co-pilot keep an eye on things and advise or even take care of some tasks while the PIC focuses on others.
Here’s something to try… next time you switch to approach control, speak in a robot-like voice… now, imagine approach responds in like manner..! I think I’d abandon the approach, pick a different airport, and use my normal voice.
Hand flying airplanes is not as smooth as that done by automation is true. But why and how is another scenario. In modern day jet flying, autopilot is engaged as soon as the undercarriage is up and airplane is climbing hundred feet above ground. At the destination too, they are mostly auto-landing. So much of dependency syndrome on the automation has developed over the years that pilots have more or less lost faith in their beliefs and capability to hand fly an airplane if situation demands.
If you always expect something is being done and taken care of by a system, if believe less of your intuition. You lose the basic skills.
Suresh, you are quite wrong! As a 767 wide-body Captain I can tell you that the A/P is not turned on at gear-retraction. It is turned on “as conditions warrant and at pilot discretion.” Some pilots fly until the high 20’s. Some put it on at 5000′. The number of auto landings is miniscule. Autolandings only occur when required for weather or asked for by Mx. Pilots love to fly and hence make almost ALL landings as hand flown landings. I do not know your experience level but you are dead wrong!! If you do not know of what you speak…..don’t speak!
John
Now if we could only get Biden to take your advice….
I think there was a misprint: “calculate an intercept and level-off limited to no more than a few hundreds of a G. . . . “. The correct phrase would have been HUNDRETHS of a G ???
You’re right, James. It’s a typo. Should be as you say, or expressed as 100ths of a G. Very little acceleration.
Thanks,
Mac Mc
Fun fact:
I have 10khrs+ on Airbus FBW aircraft. Always loved to hand-fly these, mostly with manual trust control.
Now, I‘m flying a (North American built) airplane designed to give the pilot a proper „feel“ of the aircraft. I don‘t like it and I always engage the AP at the minimum possible height after T/O and only switch it off when established on approach speed on final…
I’m an instrument-rated private pilot with a few thousand hours. My 1979 TR-182 has steam gauges, a KNS-80 (remember those???) an Aspen PFD, an STEC-30 A/P and a GNS430. Not quite fully automated but automation can do an awful lot of the flying for me and I really got to liking it.
But I have found that I still need to hand fly the airplane just about every time I get in the air, in part because I’m not flying often enough to maintain hand-flying proficiency if I let the automation to all the work. It’s not just me, either. I can’t cite the reference but apparently there have been a sizable number of ASRS reports in which pilots seem to think a lack of lfying time during the pandemic has caused their flying skills and muscle memory to degrade.
I’m no longer an active pilot but I’ve experienced the situation described here. In 2005 I was introduced to the G1000 and I decided it had a big advantage over steam gauges after an instrument proficiency check (probably called something else now) cleared me to fly with it. I usually flew single pilot in a Cessna 182 or 172SP. Over a period of time I learned to work with flight plans and became a MIC. I would get my clearance on the ground, load it into the G1000, and take off. (This took some doing if Clearance Delivery gave me a void time in 5 minutes, which New York approach wanted to do every time.) All I had to do was manage the altitudes as ATC cleared me for any changes. If I got new routing, I asked for an “initial heading” and ATC was always willing to provide one. Then I’d put it in heading mode and while George flew, I typed in the new flight plan. I was able to look for traffic, talk to ATC, and do the other piloting chores we all deal with.
Mac didn’t mention the IPC in his article. I believe it is very important! For most GA pilots, we don’t amass the 6 approaches and other maneuvers required in 6 months in IMC so I just scheduled an IPC with a CFII every 6 months as a matter of course. In the IPC, the instructor was always willing to give me maneuvers to hand fly that I wouldn’t normally encounter in IFR flights, such as holds. I turned the flight director and autopilot off. If desired, I could have limited myself to VOR navigation — the G1000 can certainly process VOR input. Of course you can still look at the moving map so you have more situational awareness than you would with steam gauges, but at least you won’t lose your hand flying skills in just 6 months.
I think that, as a species, as we evolve into a silicon and titanium based lifeform, the problem will solve itself. But it will be a fun discussion for the intervening millenia.
I am blessed to fly a 1986, A36 Bonanza with GTN 750, GTXi500 and GFC 600 autopilot. I typically turn the autopilot on at 800-1000 ft AGL and don’t turn it off until final approach inbound, sometimes at final approach fix with gear extension. It is unbelievable the decrease in workload. I try to fly with an instructor 3-4 times a year to keep the ability to do what the autopilot does for me. I can’t but help to feel that I am less proficient but don’t want to go back to the “good old days.” This is the dilemma of current technology.
I usually fly an old 182 RG with steam guages. Yes, it has an auto pilot. No, i don’t use it. I have turned it on from time to time to make sure it does indeed still work, but then i promptly turn it off. I am pretty much just a day VFR pilot at this point. I have never flown a glass cockpit aircraft. Sat in one once on the ground, but that was it. About 10 years ago, another FBO on our field approached me as i was getting out of an old 172. He tried to steer me away from my current rental agency, into his, with their shinny new 172’s. All sporting glass cockpit G1000 panels. I asked about the costs, and he quoted me $50.00 an hour more than what i was paying, for his machines. I shut him down BIG time. I said your 172’s don’t fly any faster, carry any more weight, have any more room in them, than what i just climbed out of. He turned and walked away from me. Then i got to thinking. That $50.00 an hour rental fee he was asking for in a 172 glass cockpit, i could put towards flying a steam guage 182 RG. Which i indeed eventually did. And i have never looked back. I am all for technology, don’t get me wrong. But Jay Leno said it best, and i quote: The best of old technology is better than the best of new technology. End quote.
Mac, yet another interesting and thought provoking piece of aviation journalism. I think that I enjoy your articles, so much, because you speak from real world experiences so similar to my own. I am lucky enough to have been type rated in 4 different jets. Two of those requiring a crew of two and two of them being single pilot. Two of those type ratings were done in the airplanes and two of them at CAE Dallas. In all 4 the instructors were adamant that we use the automation as much as possible. My suspicion is that this was the most sure path to a successful check ride and the most expedient technique for the instructors to achieve their number one goal, a successful student out the door. In none of those cases were we counseled to mix hand flying with automation in our real world flying. All of my type rating check rides were a cake walk using the automation as much as allowed. However, I vividly remember my first Beechjet recurrent sim training debacle. After a year of engaging the AP at 500′ after takeoff and and disengaging at below 1000′ on approach, my hand flying skills had deteriorated and I embarrassingly stumbled through my 4 days of recurrent sim training. After that humiliating experience I immediately started hand flying in almost all situations below 10,000′. And, yes at first it was definitely not as nice a ride for the passengers as that provided by the super smooth AP. That was with a required crew of two so it quickly became really easy as the pilot not flying actually did most of the work. Now, I fly a Beech Premier jet single pilot. That is a little bit trickier in that hand flying and complying with ATC instructions in a busy terminal area sometimes leads to a less than perfect ride for the passengers. I am, however, lucky enough to fly about 1/3 of my flights empty either on the way to pick up passengers or headed home after dropping same. On those flights I typically hand fly below 18,000′. In all cases the hand flying has really helped keep me sharp on the controls and eliminated any repeats of that embarrassing recurrent experience.
Mac, great subject! I am a “dinosaur”. I have flown a lot of aircraft thru the years from DH Herons, to B747’s. Until everything was lumped together (flt director, nav, auto throttles, and yes autopilots) most of us stayed pretty proficient at hand flying, and I will tell you it was better than the a/p could do.
I became a flight instructor on DC-10’s at Pan Am after the merger with National Airlines. Our 10’s had the latest R-nav from Honeywell and the autopilot was fantastic, so much that no one wanted to hand fly. UNTIL the a/p started kicking off and people realized their hand flying skills had gotten away.
Several guys actually bought GA aircraft to keep their hand flying skills up. So much so they took the cost off their taxes as an expense needed for their job. Not to mention they wanted an airplane.
Later at another company while flying the B-737’s, we went from “steam” to modified glass, meaning the ADI and HSI were glass. Not a problem, then we got 737-800’s, full glass. With it a blended category, the company worried about the pilots going from Steam to Glass. Wrong, it was going back not forward. Glass, is awesome. However, when it doesn’t, is the answer going to be push button and let Garmin “Take you home”?
Boeing has gone through a nightmare with the MAX because countries with inexperienced and pilots that weren’t allowed to hand fly, is it any surprise they could not HAND FLY. They didn’t know how because they put the a/p on after gear retraction and left it on till landing.
Let us not forget, Asiana B-777 crash in SFO in 2013 that was caused because the autoflight disengaged and they couldn’t HANDFLY, they didn’t even know what to do with the throttles.
If you are going to go up in the air, know how to fly. Use to be said “if you can fly a Cub, you can fly a 747, however the opposite is not true”
I am so glad to have learned how to fly in tail draggers and round motors.
There are Pilots then there are people that fly….. there is a difference.
+