A ‘robot pilot’ has landed a Boeing 737 in a simulator

“Ladies and gentlemen, this is your captain speaking.”

What if those words so familiar to air travelers weren’t spoken by a human pilot, but by a robot?

The US military is developing technology that uses artificial intelligence (AI) and robotics to create machines capable of flying all types of aircraft, including commercial airliners.

A robotic co-pilot developed under the Defense Advanced Research Projects Agency’s (DARPA) ALIAS (Aircrew Labor In-cockpit Automation System) programme has already flown a light aircraft.

And now it has flown and landed a simulated Boeing 737, show in the video below.

Beyond autopilot

Some may wonder what the difference is between this new technology and the autopilot systems that can fly commercial airliners between takeoff and landing. Or indeed the pilotless drones that are regularly used by the military.

DARPA says its aim is for ALIAS is to create technology that supports “execution of an entire mission from takeoff to landing, even in the face of contingency events such as aircraft system failures”.

To do this, the technology currently being developed in the ALIAS programme by Aurora Flight Sciences moves beyond computer software and uses a system of cameras to enable the machine to “see” what is happening in the cockpit. It can read all the cockpit instruments and gauges and adjusts them using a robotic arm.

Another key aim of the ALIAS programme is to develop technology that can be used across many different types of aircraft, including older models. This is important, says DARPA, as avionics and software upgrades “can run into the tens of millions of dollars per aircraft”.

DARPA hopes the ALIAS programme will produce an automated system that will be cost effective because it can be easily transferred from one type of aircraft to another.

In addition to the 737 simulator and the Cessna light aircraft mentioned above, Aurora’s system has also flown a Diamond DA42 light aircraft and a Bell UH-1 helicopter.

Using machine learning, the system learns how an aircraft handles – not only from being onboard, but from a database of hundreds of thousands of previous flights.

The combination of its learning capability and robotic arm should also mean the machine can act much more like a human pilot and deal with unforeseen circumstances better than existing autopilot software.

What next?

At present, DARPA is developing the technology in order to address a shortage of qualified pilots in the military, hoping to deploy the automated systems as co-pilots.

Rather than replacing pilots, DARPA officials told Aviation Week that ALIAS would mean “pilots can use their time more productively” by allowing them “to think about the context of the mission and the information being collected, versus managing the stick and throttle”.

It is expected that elements of the ALIAS technology could be adopted within the next five years, similar to the way that car makers are gradually introducing elements of automation into the motor vehicle market.

However, in the longer term, just as developments in the motor industry point towards a future with automated vehicles, some envisage a day when humans flying aircraft will be a thing of the past.