Boeing said it’s building an airliner intended for only one pilot. Is that a good idea?
In short: not yet.
The technology itself is essentially there but not at the reliability that we need. It’s about 90% there but there’s a lot of devil in the last 10% of detail. And there are other problems.
Security: Murder / Suicide
Remember Germanwings? Where the copilot committed suicide and took everyone with him? Yeah. After that it became a requirement to have someone else in the cockpit when one crewmember exits. It was already existed in most countries for cockpit security.
There is still some risk of a rogue pilot but in less likely ways: FFDO’s (one pilot has a gun) and low-altitude flight diversions (one pilot pushes over on short final). There’s not much we could do about these risks, at least for some years, until automation can be counted on enough to prevent pilots from intentionally crashing.
Presumably the risk from FFDOs is reduced because they go through extra psychological screening. And remember, in many billions of miles flown, neither of these has ever happened. The odds are with us.
Complex Environments
There are still a lot of variables that would be excruciatingly difficult to plan for, like when ice, thunderstorms, malfunctions, or other unexpected problems come up.
When weird stuff happens the human is better equipped to handle it, but having two humans is far better to bounce ideas off of and to monitor. If one pilot is actively controlling the plane, he loses wide focus on surrounding needs. Enter the “pilot monitoring.” If automation has checked out, or is compromised, it will be a stressful time that gets handled much better with two pilots.
Using Automation Properly
Automation has helped make airline flying the safest way to move humans horizontally ever devised. Safer than cars, trains, bikes, walking, busses, and boats. Elevators may be safer per mile but those move vertically.
Even so, there’s still room for improvement.
Bubbles
Immense safety improvements still lurk, starting with better warnings as described here.
- Have a spatial bubble around the airplane. It knows where it is, where the runway is, which runway its going to, and where the terrain is. Sing out of something is amiss. We already have some of this with EGPWS but there’s no logic about runways.
- Have a performance bubble where the current performance is weighed against what’s needed and sing out when the two don’t match. If you’re accelerating down a runway, for example, and the system recognizes you won’t make it based on the the current flap setting, acceleration, runway and terrain, sing out well before the abort decision point.
- Hav a temporal bubble around the airplane. If some parameter will be out of bounds within 5 or whatever seconds, sing out. For example, if the airspeed is getting slow and the throttles aren’t pushed up in time, sing out early. “Airspeed Low.” Right now that warning comes unnecessarily late. Use maneuvering speeds/AoA’s at current flap settings instead of just stall value. Yes, I know, airplanes stall at the critical AoA but pilots fly at set airspeeds, not AoA’s, so include that parameter.
How about landing on the wrong runway, or worse yet, the wrong airport. Yup, that still happens and has come close to killing a planeload of people. The plane knows where it’s landing so it would be trivial to warn the pilots. If they get below 1000′ and they’re not within 3 miles of the CORRECT runway, sing out: “CHECK RUNWAY!”
Airline policy would require action in like how it requires an escape maneuver upon hearing “TERRAIN, PULL UP.”
Pilot Preparedness
If a physical/mental skill is not used it rots. That’s happening. We rarely fly airspeed/heading/altitude manually anymore, especially in demanding situations. That’s a bad plan since solid airmanship be be required in emergency situations where automation is unavailable. If we expect pilots to be valid backups, their hard-won skills must be maintained.
Airplane
We can indeed get some of this flying the actual airplane with passengers on board. But frequently, even “hand flying” still doesn’t involve everything since autothrottles are commonly left on, and the other pilot is running altitude and heading selectors by airline policy.
Simulator
This is where basic airmanship should be practiced to a high degree. Have pilots controlling heading, airspeed, altitude precisely in various configurations. Of course pilots don’t want extra sim sessions nor do they want their livelihood to have any more breaking points. I don’t.
How about a computer generated score that has no effect on pass/fail? The score is reduced by diversions from parameters. Configuration changes are REPEATED, including go-arounds, starting with the most common ones, namely going from a stable glideslope descent to a go around.
Cost of Improvements
If every airline was required to do these upgrades and flight training then the fact that every ticket would cost an additional 45 cents would have no appreciable effect on travel. We do have to be mindful of costing people into their far-more-dangerous cars.
Same for other safety improvements. We do want to pick the low hanging fruit, the cheapest cost per life saved, but there’s a lot of that out there.
See also the cost per life saved.