AI in Aerospace: Boeing’s Helen Lee
24:10 Share

Today, we're airing an episode produced by our friends at the Modern CTO Podcast, who were kind enough to have me on recently as a guest. We talked about the rise of generative AI, what it means to be successful with technology, and some considerations for leaders to think about as they shepherd technology implementation efforts. Find the Modern CTO Podcast on Apple Podcast, Spotify, or wherever you get your podcast.

Many tasks could benefit from advanced technologies, but how can organizations use emerging technologies in high-stakes situations? Find out how the aerospace industry is approaching AI in today's episode. I'm Helen Lee from Boeing, and you're listening to Me, Myself, and AI. Welcome to Me, Myself, and AI, a podcast on artificial intelligence and business. Each episode, we introduce you to someone innovating with AI. I'm Sam Ransbotham, professor of analytics at Boston College.

I'm also the AI and Business Strategy Guest Editor at MIT Sloan Management Review.

And I'm Sherwin Kodabande, senior partner with BCG, and I co-lead BCG's AI practice in North America. Together, MIT SMR and BCG have been researching and publishing on AI for six years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities and really transform the way organizations operate.

Shervin and I are excited to be talking today with Helen Li, Regional Director of Air Traffic Management and Airport Programs in China for the Boeing Company. Helen, thanks for taking the time to talk with us. Welcome. Thank you for having me. Let's get started. Helen, can you tell us about your current role at Boeing? I currently work at Boeing China in the Beijing office.

My main responsibility is to oversee Boeing's airport and air traffic management programs in China. And we try to do here is to improve the airport and airspace operational efficiency and at the same time to enhance flight safety.

I've been doing consulting for airport and ATM for 10 years. And the interesting part is every project is very different and able to work with different group of people. Air traffic management is not the core business for Boeing, but what we really want is to help our customer.

And because airlines are our customer, we want them to operate more efficiently and also enhance their flight safety. Other than design and manufacture of aircraft, we also provide a lot of services to our airline customers. It's kind of like an entire ecosystem that Boeing and actually other partners, we're working together on it. For example, we provide flight planning services.

to airline customers. And so that will help them to better plan their flights and optimize their flight path. And that also will reduce their fuel consumption and emission of carbon dioxide. Air traffic management has always been quite fascinating to me because I always think of there's tens of thousands of aircraft at any given time in the air. And

You have different systems trying to do air traffic management or air traffic control for that aircraft. It feels like it's very data-centric. It's also somewhat maybe chaotic where unpredictable things can happen. And then you probably have, I don't know, you tell us, some interaction effects between different ATMs or air traffic management systems. But can you just educate us and our audience a bit

on what it is and how it works and how sophisticated and complicated it is? So you probably all know that a fly usually take six phases. Tessie on the ground, either tessie in or tessie out, and then take off from the runway and then start climbing and then en route and then start descend and land.

So there are six phases of it. And in each phase, there are air traffic controllers controlling the aircraft. General aviation in the U.S. is very different. It could fly in controlled airspace. But for most of our commercial flights, you need to go through these six phases first.

First, you have a controller at the ground, and you probably see that airport at the high tower, that's the ATC tower. They have the ground controller that controls the ground movement, and then there's a tower controller that controls the takeoff and landing. And once the aircraft climbs to a certain altitude, it will be handed over to the approach controller.

or departure controller, you can say that. So that controller will handle the aircraft to about 30 to 60 nautical miles from the airport. And then you are passed to enroute controller. And then the enroute controller usually have to go through several centers. And in each centers, there could be many sectors. And each sector will usually manage by one controller.

And so they hand off one by one all the way through to the landing site. So that's how they work. But I would have to say AI is still not widely used in ATM system. There's many, many challenges. We can talk about that later. But they're still using voice communication mostly. And so that's the part that

haven't been replaced by just data link communication yet, but we're moving to that direction. One application that's widely being used is computer vision, like image recognition. And so, for example, what we are doing right now is one of the studies we try to do is to use that to recognize weight turbulence.

the weight vertex right after like a landing aircraft, maybe an aircraft approach. And so we're using the LATA machine to observe those weight and then we use the AI algorithm to help us to capture those. And so we can train the machine and able to recognize the location and the strength of that weight vertex.

That would be something we might be applying in the future to shorten the way turbulent separation. And another one is, like I mentioned earlier about speech recognition. That is something that we are doing a lot of research on and not just Boeing, but other parts in the industry to speech recognize the conversation with controllers.

And because some of the instructions from controller is kind of the same from one aircraft to another aircraft. And so that part may be able to digitize and just use a data form and have a display in the cockpit instead of having the controller to repeat it all the time. And the other one, the aircraft have to be rerouted to another aircraft.

path and that will be another thing that can be digitized using the speech recognition. The benefit of it is if it's all digitized, it's all come from the controller and then we don't need the pilot to punch in the path, the waypoints into the system and it can be uploaded directly to the flight management system. So they will avoid some of the errors made by the pilot.

So we're hearing there's a lot of potential applications, but not yet widespread use. What are some of the reasons for that? You said voice recognition, for example, right? Yes. Like my...

I mean, Siri doesn't have any problems recognizing my voice, although I have heard on YouTube some of the conversations between air traffic control pilots, and it's as strange to me as deciphering my doctor's prescription. But tell us more, why is that hard? One is the reliability. You know, how reliable when they recognize the voice, because it cannot make any mistake.

In air traffic management world, it's zero tolerance for mistake. So it's not like your theory that if they make mistake, that's fine. You just say again, right? But not here that you cannot do it.

And so that's the one thing. And the other one that I would say probably with data link communication is the security, whether we have a very secure environment that nobody would hack in or things like that. That's why so far we haven't seen any application being certified yet. If I'm understanding you correctly, one of the biggest hurdles is the need for

absolute precision, zero error tolerance, and just how much is at stake that it's maybe unlike many other things. Yeah, Shervin, I think a lot that we talk about is corporate application where people make a recommendation or a loan approval. It's not real time. It's not critical in the moment.

Whereas this is a very different scenario. But not all the scenarios are very different. Like, for example, the wave turbulence is something that doesn't have to be real-time. That could be an after-the-fact analysis. Runway configuration could be after-the-fact. And it's not just real-time either. It's like if you think about algorithmic trading that's going on, it's near real-time or the credit card authorization is real-time. I think it's real-time, but also...

how much is at stake, like the cost of being wrong, right? Like image recognition, for example, or video, like in medical applications, you still have a doctor. And if there's a mistake, God forbid, it's one life,

It's not hundreds of lives. But are we getting it that it's really the gravity of the situation is that what prevents these things from being widely adopted in the flight life cycle? Actually, in the last couple of decades, the industry is really preparing for this technology to be applied in this industry. So a lot of work has been done to improve

make the automation of the system. So that's one part. Now we know most of the aircraft, especially new aircraft, the advantage is to use this in China is they have almost all the new aircraft in China because most of their aircraft less than 10 years old. So that means they all equipped with the latest technology on board. And so that's one thing. And also their control system

they use the most advanced technology as well. I would say the basic for an AI to be applied is to have some of the automation system to be there. One thing that we are also doing, you know, aircraft can do auto-fly.

Before, many times we used the procedure coming, for example, land and depart from an airport. But now we are really promoting the performance-based navigation. And since most of aircraft already have that navigation system equipped, we're able to give aircraft a more precise route for it to climb out or descend to an airport.

And that means it will be much easier later on if we try to manage those aircraft. So that's one thing, kind of like create a base so we can build upon it and to use more advanced technology in this system. And there's many new studies coming out and there are roadmaps and plans for using AI in air traffic management.

So I would say in the next decade, we probably see a lot more things come up using AI. I would have to say another challenge we have in using AI is usually the AI, if you apply to air traffic management system, that we might rely on a knowledge-based expert system. So it's very hard to build a good expert system.

Especially, you know, different environment, their expert system may be completely different because their operation is different. They may have a terrain, they may have different runway configuration and all that. So that's another part is you cannot just build one expert system to use for everywhere. So there's a lot of preparation work. And what struck me as particularly interesting about what you're saying is

is how coordinated that needs to be with lots of different people, lots of different organizations, different airlines. This isn't just a thing that one organization can put in place and dictate to their people that they use. It's something that has to coordinate across lots of different organizations with equipment like airplanes. You can't just, oh yeah, well, let's just all get new airplanes next week so they have the new technology. It's a complicated process.

They're very right about that. And that's why everybody's working on what we call SWIM. That's the System-Wide Information Management.

So that means we're able to share information between different players. That could be airline, could be pilots in the cockpit, could be air traffic controllers. And then we will have the meteorology data and everything will come together. So everything will be shared within the system. So different players will be able to see the information and data they need to better operate their own systems.

So you pointed to a lot of forward-looking aspects of artificial intelligence. Is there something you're using right now? Is there something that Boeing is doing right now that maybe we don't know about or that is behind the scenes, it's hard for people to see? What kinds of artificial intelligence are currently in use right now? One thing that is coming very close in application is using image recognition technology.

I listened to one of your previous sessions from the lady from DHL. She mentioned a similar technology. So for example, when the aircraft coming in, we can use a robot camera to take pictures of the aircraft, to take pictures of the fuel sludge to see if there's any damage, and then use AI to recognize whether there's an important thing that we need to take care of, whether you need to go into the hangar to be fixed.

you know, things like that. That seems like a great application. Yeah, but before, you know, you have to have a human being to walk around the aircraft to identify all those and then make decision. I think what's difficult about some of those things is

What you don't notice is when they don't happen. Let's say you do a great job of inspecting the plane beforehand and finding a problem and preventing it, or recognizing a part needs service before the people are actually on the plane. These are not things that people notice. You only notice when it doesn't work. Correct. It's a classic engineering problem. Yeah, absolutely right. We're in the process to collect data because we need a lot of data to train the machine.

And the most important part is to collect all those data. And nowadays, the new aircraft, like the 787, there's a lot of data we can collect. It's not like the older 747 that was built decades ago. But the new aircraft that we make today, we're able to have a lot of data. And then those data will help us to analyze the health of the aircraft.

That seems great. So, Helen, we have a new segment and we ask our guests a series of rapid fire questions. So just answer the first response that comes to your mind. You don't have to think about it too much. Just your your first reaction. So what's been your proudest moment of using artificial intelligence?

It's hard because I don't use that every day. Well, I think that was the response. Exactly. That may be your answer. No response is the response. Okay. What worries you about artificial intelligence? The challenge we were talking about. You know, how safe it can be if you really apply to the air traffic control environment. What's your favorite activity that involves no technology?

Oh, that's something I couldn't do it now. I used to do kayaking, but when I was living in Atlanta. Oh, okay. Well, I'm actually from Atlanta. So yeah, we probably kayak the same waters then. I see a picture of you snowboarding on your background there. Yeah, that's in Beijing. I used to do snowboarding when I was living in DC area. So what was your first career that you wanted when you were a child? I would have to say

mechanic engineer because that was more my mom been doing because she designed household electrical appliances I used to watch her draw engineering drawings when I was little so I thought oh that's amazing and then you can see the product and that's fun

So that's why my major in college was mechanical engineering before I changed to aerospace engineering. Well, both Shervin and I are chemical engineers, and so I have to bring that up every time. So we believe in chemical engineering is better than all the other engineering. Well, yeah, in some of the universities, aerospace engineering is part of the mechanical engineering department. So what's your greatest wish for AI in the future? A wish?

All the aircraft will be able to control by AI, and also the air traffic control will be conducted by AI. So it's always just machine-to-machine talk. And so there are less error, less chance for mistake, and we're more efficient. I think we all want those things. Helen, it was great meeting you and talking with you. I think one thing that impressed me about this is what a complex environment you were in

coordinating lots of different organizations with equipment that's really out of your control. It's a very difficult situation compared to a lot of the people that we talk to. Thank you for taking the time to talk with us. We really enjoyed it. Thanks. Thank you very much. It's been really enlightening. Thank you for having me. Next time, Shervin and I talk with Somya Kotipati, Vice President, Global Supply Chain Technologies at the Estee Lauder Company. We hope you can join us.

Thanks for listening to Me, Myself, and AI. We believe, like you, that the conversation about AI implementation doesn't start and stop with this podcast. That's why we've created a group on LinkedIn specifically for listeners like you. It's called AI for Leaders. And if you join us, you can chat with show creators and hosts, ask your own questions, share your insights,

and gain access to valuable resources about AI implementation from MIT SMR and BCG, you can access it by visiting mitsmr.com forward slash AI for Leaders. We'll put that link in the show notes, and we hope to see you there.