Just how smart can an airplane be?

One day, smart enough that people will be able to get around in on-demand self-flying taxis, predicts Irene Gregory, senior technologist for advanced control theory and application at the National Aeronautics and Space Administration.

Today, planes are very smart, but humans still call the shots every step of the way. The technologies to turn science fiction into reality are starting to emerge, though, bringing the next big step into sight: planes that can think for themselves, to operate more efficiently and improve their safety in flight.

“Think of George Jetson and his car,” said Dr. Gregory during a presentation at an American Institute of Aeronautics and Astronautics conference in Washington, D.C., in June. She was referring to the character from “The Jetsons,” the animated TV comedy from the 1960s that portrayed a future where George commuted to work in a flying car.

That’s not precisely the future Dr. Gregory envisions, but it gives the flavor of the flying taxis she does expect to operate someday alongside piloted airliners. It’s a future in which the machines, not the humans, are in primary control of flight. “You don’t want to touch anything, you want to tell it ‘I want to go here’ ” and have the aircraft take you there, she said at the conference.

The working goal, Dr. Gregory said, is to design a “self-aware” plane, able to monitor the condition of the plane and its operation as well as what’s going on in the environment around it—and then tie all that information together to make decisions about how to operate the plane safely.

A self-aware aircraft would be able to adjust its flight plan to respond to problems and anticipate new ones ahead, ranging from technical malfunctions onboard to adverse weather and emergency situations.

Many of the technologies that might one day add up to a self-aware plane are already in use today. The latest airliners from Boeing Co. and Airbus Group SE, for instance, have sensors that measure thousands of safety parameters inside and outside the planes.

But the self-aware aircraft would still be a major leap beyond any technology in today’s planes, including autopilot. With autopilot, pilots can set a plane’s flight path so that they don’t have to guide the plane themselves. But the plane will follow that path no matter what happens, until a pilot changes the course on autopilot or takes over the controls. A self-aware aircraft would be able to change course, without any human intervention, if it determines the plane is in danger or another route would be more efficient.

Self-aware aircraft also would be an improvement on the equipment airliners already carry that tells pilots how to avoid midair collisions or to prevent flying their planes into the ground. It would enable planes to take these steps themselves. In cases where a plane is being flown by a single pilot, that would reduce the pilot’s workload and increase safety. And it would be a fundamental requirement for unmanned aircraft.

Drones are pioneers on this front. Regulators and engineers are attempting to find ways to manage increasing demand for drone services without endangering commercial or private flying. That’s caused them to explore systems that allow drones to “sense” other planes nearby and take action to avoid them.

Self-aware planes also would consider factors beyond flight operations and external conditions. In the same way humans know to go easy on a sprained ankle or sore back, the self-aware aircraft would be able to understand how the health and history of its structural skeleton would affect how it could perform in flight, Dr. Gregory said in her presentation. For instance, the plane would know what types of stresses it could endure if it needed to make a sudden maneuver.

That understanding would be enabled by existing technology called a digital-twin software simulation model, which combines manufacturing data for a system with continuously updated data from operations to predict when structural parts might fail or need maintenance. The simulation, in human terms, would tell you how fast you can run today and when you’re going to need your knee replaced. The U.S. Air Force is trying to apply digital-twin simulations to better predict the costs and capabilities of the weapon systems it buys, and General Electric Co. wants to use digital-twin technology to improve its wind turbines and jet engines.

Challenges abound

The development of self-aware planes still faces several challenges, including the need for continued advances in the development of a digital brain—known as an adaptive mission manager—that can be relied on to make sound decisions on par with or better than humans.

Then there are more-mundane challenges, such as the danger that too much information will gum up the works. A common problem today is oversensitive sensors that sometimes result in software issuing warnings to flight crews during preparation for takeoff about, say, a problem with a plane’s flaps when there’s nothing actually wrong. One concern is that more-complex sensors could result in pilots being flooded with spurious error messages.

Today, such warnings can be dealt with relatively quickly. The systems being monitored are designed for easy access so that a maintenance technician can check them for problems. But error messages from sensors embedded deep in a plane’s structure could be more difficult to resolve, and that could slow the whole process of maintenance and departure.

To get to highly automated flying, the industry will also have to overcome the doubts of consumers and regulators. For a sense of the difficulties ahead on that front, consider that car makers are still struggling to convince regulators to allow autonomous cars on the road. And that challenge became potentially more difficult after the May 7 crash of a Tesla Model S car caused the first known fatality while a vehicle was in autopilot mode.

But the prospects may be better for intermediate steps that could offer a safety net for pilots. Several high-profile crashes in recent years have been linked to pilots losing control of their plane because of mistakes they made in responding to in-flight emergencies.

NASA’s Langley Research Center in 2010 and 2011 flight-tested a 5.5%-scale airliner, repeatedly trying to stall the aircraft, which would cause it to lose lift and make it unflyable. The digital system, jointly developed with Boeing, safely recovered the aircraft. In practice, if an airplane sensed the pilot was losing control and it was reaching an unflyable state, the system is designed to automatically augment the pilot’s control and recover stability.

Dr. Gregory says in an interview that the system could be put into operation on the next generation of all-new airliners, assuming it wins the approval of the Federal Aviation Administration. The FAA declined to comment.