When flying between two cities, aircraft may not necessarily take the same path. Instead of traveling between New York and Seattle on the same, constant route, each voyage must follow a flight plan that takes it over particular waypoints. While you may believe that the plane’s captain makes the decisions, this is the job of those who work for the airline on the ground. They’re known as dispatchers.
“The dispatcher and the captain share legal responsibility for every flight that an airline operates,” says Pasha Saleh, an Alaska Airlines pilot who also oversees the company’s innovation and flight operations strategy. That dispatcher devises a plan, submits it to the FAA, and then sends a cleared plan to the pilot so he or she knows where to go.
The issue is that human dispatchers must examine a large number of complicated factors. They’ll be thinking about things like weather, wind, and military airspace that are off-limits. “They have to go through these ancient archives to collect all this information—in certain situations, they have to go to an FAA website and decode alphanumeric sequences of text,” Saleh adds. “In the background, the Weather Channel is constantly on; CNN is constantly on.”
All of this information is inhaled by the dispatcher, who then spits out a flight plan. They may either design their route, which takes time, or they may pick an off-the-shelf alternative, such as route three. That is a more efficient method. “They’re chopping with an axe,” Saleh explains. After all, creating a bespoke route with a scalpel takes much too long.
Artificial intelligence excels at quickly crunching large amounts of data, and new technology at Alaska Airlines from a firm called Airspace Intelligence is coming up with speedier routes. Human dispatchers remain in the loop and have the option of using the AI-generated route on a case-by-case basis. They continue to submit it for approval to the FAA.
Saleh explains, “They’re all custom routes today.” “They’re pathways that have never existed before in many situations, and it would take a person 30, 40 minutes to think of them.”
According to Alaska, AI routes are around 5.3 minutes quicker on average than human-designed routes. ETA estimations are also improving. The technology was officially launched in March, although the corporation had been testing it for six months in 2020. During the testing period, the business claims that the system saved roughly half a million gallons of gasoline, saving 4,600 tonnes of carbon from entering the environment.
The program, known as Flyways AI, identified a technique to optimize a flight’s path 64 percent of the time throughout the experiment. Human dispatchers choose to utilize the resultant route roughly 30% of the time, a figure that has remained consistent now that the system is in regular operation. In a nutshell, it’s similar to Waze advising you on the best route from point A to point B, which you may or may not trust enough to follow.
The Waze analogy works well since aviation software is similarly seeking to minimize delays – a flight that may ordinarily be on time, but is delayed at the last minute due to a congested arrival airport. Wind and weather are aspects that both humans and AI assess, but “what the AI adds is the purpose of every other flight in the national airspace system,” according to Saleh. That information is “technically available to a human dispatcher, but impossible to incorporate.”
The device can also see into the future. “The AI is thinking, ‘This is what’s going to happen six hours into that journey, so let’s start the trip off in a different route to avoid anything later on,’” Saleh explains.
Importantly, this new software does not replace Alaska’s traditional system for submitting flight plans to the FAA. That old flight planning engine is still in use. In certain circumstances, the AI simply offers a new route, which the dispatcher approves before sending it through for approval. That’s a good thing because removing the core software would be a major undertaking. He points out that replacing the flight planning engine would be like doing open-heart surgery on the airline.
In that sense, it’s similar to having the Waze app installed on your phone. Of course, the software on your smartphone does not communicate with the code in your automobile. The systems are distinct, and the choices are still made by the human driver. In this scenario, an AI assistant assists the dispatcher in routing an airplane more effectively some of the time, allowing passengers to get to