What if technology could help pilots recover an airplane when it is clear (to the software) that the pilot’s actions are trending toward an accident?
Consider the Air France 447 and Colgan Air 3407 accidents. In the Air France accident, the Airbus A330 stalled into the ocean from cruise altitude, following icing of the pitot sensors. The consensus is that the pilots just needed to move the control stick forward to unload the wing and the Airbus would have resumed flying. In the Colgan accident, the pilots failed to add power after their Dash 8-400 slowed while leveling off during approach. The Dash 8 was on autopilot, and the autopilot kicked off when it reached a limit, handing the pilots an airplane that was about to go out of control. Investigators concluded that the captain reacted incorrectly, and the Dash 8 stalled and crashed.
In both cases, the pilots were not ready for what happened, and they did not figure out what was wrong in time to prevent an accident. Also, it is assumed that the airplane designers didn’t anticipate those occurrences or the way the pilots reacted. No designer can anticipate every way that a system or a pilot can mess up. But what if technology could help the pilot rescue the situation when it is clear (to the software) that the pilot isn’t about to resolve the problem? If this is even possible, does it make sense? And, what would something like this look like in practice?
To answer these questions, AIN interviewed avionics manufacturers and asked how their products can help pilots and also what developments in their back-room laboratories are under way to address these problems.