Are You Ready for Only One Pilot in Your Cockpit?: Hal, Is That You?

Originally Published on June 29, 2018

What it means to “pilot” a commercial airplane has shifted over the years, and a current proposal is challenging the unique relationship pilots have with their airplanes.

Let’s begin with what hasn’t changed. Most pilots begin by learning raw airmanship skills in basic training airplanes. They spend hundreds of hours flying an airplane with the most basic systems and no automation. As they move through the hours into bigger, faster, more complex airplanes, the airplanes actually get easier to fly (but don’t tell anyone that) because of automation.

Aviation’s incredible safety record in recent years is evidence that the training and safety culture is working. There are exceptions to every rule, but given the infinite variables of aviation and the billions of miles flown, flying is by far the safest mode of transportation. The industry has learned through trial and error and never stops learning, and trying, new technology to increase safety. So, it was inevitable that someone (probably an accountant) would look at the current pilot shortage and say, “Hey, why don’t we just build an airplane that only needs one pilot to fly it? Or, better yet, let’s replace both pilots with a computer!” Gasp.

During the glory days of aviation, there were often three pilots in the cockpit – or at least two pilots and a mechanic. I flew the B727-200 in all three seats and know the magic of a synchronized flight crew that knew the airplane inside and out and it was not just three times safer, it was infinitely safer. And I also know the dynamics of a crew when one pilot believes CRM is “Control Reality Myself.” Having one pilot in the crew not willing to work as a “crew”, or doesn’t understand the systems thoroughly, can be exponentially more unsafe.

When the Queen of the Sky, the Boeing 747, was converted to a two-pilot crew, the aviation industry and pilots were insulted to think a pilot could be replaced by a computer. Secretly, we knew that the flight engineer could be replaced, but who was going to get us our sodas when the flight attendants were busy or do the exterior preflight when it was raining?

Artificial Intelligence has come a long way since “Hal” took control of the spaceship. We can all giggle now when we ask Siri “Do you have a boyfriend?” and ‘she’ says, “No, but drones are always trying to pick me up.” But we can also marvel that we have a rover on Mars, or that there are self-driving cars that can parallel park, or space crafts that can take a payload to space and return the rockets that sent it up. With Performance Based Navigation, we can cross continents and oceans and at the end of the flight, the GPS and autopilot is so accurate, down to the meter, that it can land an airplane all by itself in zero/zero human visibility. It makes us question and wonder what it means for a computer to think, learn and understand. Can we teach a computer to understand what it’s thinking and have its thinking corrected when something in its environment changes?

To take a complex topic down to its basic element, consider the thought process of a computer as a belief system. The programming of a computer creates a belief system for it. The computer believes it without actually having seen it because it was taught to believe, without a doubt that “this” means “that”. It cannot know anything outside those boundaries. Yet.

Last week, I opened my freezer and I thought the sky was falling as hundreds of ice cubes came crashing out. My ice maker believed it was empty because its arm that senses ice level was stuck in the down position. It believed it needed to make more ice. It was doing its job accordingly and amazingly well, it just didn’t know it had a broken arm so it kept pumping out ice cubes and filled my freezer full of ice. I swear I heard it laughing as I shrieked in surprise upon opening the door. Yes, that example is too basic to compare to a flight computer, but you get to where the concern lies in a belief.

Now, imagine an airplane crossing the ocean at night, trying to top a thunderstorm. The airplane believes its airspeed is decreasing because the pitot tube heat can’t keep up with the rate of ice accumulation so it kicks off the autopilot and starts screaming “stall”. Pilot, lacking raw basic airmanship skills who has spent his entire career believing the computer, responds by believing the information on the airspeed indicator. Second pilot in the cockpit believes the situation incorrectly as well. Senior captain comes in, figures it out, but by then it’s too late…so you can see the balance of automation interacting with humans is already tenuous, but we’re working on it. Constantly.

To now pull this all together, consider this: Every airplane you will ever step foot on has something not working, or placed, exactly the way it is supposed to. It’s not possible. Anytime you have a moving part, or a billion parts in an airplane, there are an infinite number of variables of which could never be programmed into a computer. Even if it means a bolt is not torqued within an ounce of proper applied force, there are variables. Aircraft have backup systems for everything and we consistently operate with items not working because aircraft are designed to have a circle of safety in which to operate and 99.1% of the time those variables don’t interfere with normal/safe flight operations. We already know items might/will fail, so we have systems and pilots in place to handle those issues. That’s why pilots devote their lives to understanding the variables of aviation. They are trained to think like a pilot, and to have the instincts of a pilot, so they can handle any number of variables - even if they’ve never been trained exactly on every possibility, because it’s not possible.

There are thousands of pilots who are conducting legal single-pilot operations in light business jets. These business jets are single-pilot certified and the pilots flying them conduct very safe flight operations. I spent 15 years in business aviation jets and I can fly those by myself if I wanted/had to. I could have flown the B727 by myself if I needed to, but MEL the autopilot and throw in an emergency and the odds start to turn. In aviation, there will always be a blurred line between can and should. Two well-trained pilots in the cockpit will always be exponentially safer than one, even in an airplane that is certified single-pilot...and even though there are exceptions. Add in an emergency situation on any airplane and there is no comparison.

This all leads to that .9%. That moment when an anomaly actually occurs in reality, but the computer doesn’t believe it. Right now, the basic concept of commercial aviation computers/automation is to override (or notify) stupid pilot input and keep the flight within the envelope of a safe operating profile. The computers (several of them) let the pilot know when a system is out of normal parameters. If gear and flaps are down, the computer believes you are landing…but, now let say you bounce a landing and are in this configuration while trying to climb. Same profile, different intention. Different belief.

Apr 17, 2018; Philadelphia, PA, USA; Photo from a Twitter post by Joe Marcus (@joeasaprap) on April 17, 2018 showing what appears to be the damaged engine on a Southwest Airlines airplane with a caption reading:
"What a flight! Made it!! Still here!! #southwest #flight1380" Mandatory Credit: Joe Marcus via USA TODAY NETWORK

During that .9% of emergencies, it takes the combined effort of at least two pilots to get that aircraft safely on the ground. One to fly, one to manage the system where the emergency is taking place. Sometimes several systems, like when a fan blade punctures the pressurized vessel. Engine and pressurization failure. These are two separate and distinct requirements of two pilots in the cockpit and right now, the automation/onboard computers can only tell the pilot what it believes is wrong with the airplane. It’s often correct, but only well-trained pilots will know. It’s a complicated relationship, but it works. And, as with all relationships, you must keep working on it.

I have no doubt that fully automated passenger aircraft (i.e. pilotless) are in our future. Pilots will be placed on the endangered species list someday, but not in this generation. Right now isn’t the future, but right now there is a draft FAA reauthorization bill with legislation calling for research into single-pilot cargo aircraft which has triggered this debate. I hope the “research” reveals reality. (To learn a little more, this is a good summary Flying Mag)

The aviation industry embraces any idea that makes aviation and flying safer. With our current technology and the ideas being generated from each step forward, we’re doing that. But trying to solve a pilot shortage right now by removing pilots from the cockpit for financial reasons is beyond the pale. And suggestion that pilots flying cargo, at night during the most challenging conditions, should not have the right to a full complement of safety devices, including a first officer, should raise some red flags for everyone. Why should a cargo aircraft flying over your neighborhood be operated any differently than a passenger aircraft? Single-pilot flight op lives matter too…

From the front desk of an FBO, to the captain’s seat of a commercial airliner, Erika Armstrong has experienced everything aviation has to offer. If you want to share your opinion about Hal, she can be reached at Erika@achickinthecockpit.com Erika is also the author of A CHICK IN THE COCKPIT , an aviation professor at MSU Denver, and Director of Instructional Design at Advanced Aircrew Academy.

SOURCE: https://www.linkedin.com/pulse/you-ready-only-one-pilot-your-cockpit-hal-erika-armstrong/