The next generation of engineers in the US ATM industry - Hari Iyer - ASU

Picture of Vincent Lambercy
Posted by Vincent Lambercy
 

Welcome to one more episode of Radar Contact. My name is Stojanche and today I'll be hosting this episode. Just for our listeners, it's my first time hosting episodes in Radar Contact. I'm an ATM consultant with FoxATM for a couple of months now, so I'm very happy to be here and to be host of this episode. With me today. It's Mr. Hari Iyer. He's a software engineer working in ATM industry. Hari, welcome and thank you for being here today.


Thank you. It's both a pleasure and an honor to be here and thank you for having me.


So thank you Hari. So just to begin our podcast, can you just give us a brief background summary of your professional background? Who are you, how one software engineer find the way up in the A T C industries, things like that.

 

Sure. So I hail from India ever since age four airplanes and air traffic controllers are I've thought of. And I knew that no matter what I study, this is the field that I'm going to end up in. So I went through schooling in India. I did high school and got my undergraduate in computer engineering focusing on computer science and artificial intelligence from the University of Mumbai. And during my undergrad experience, I built a project called Flight Router that was a weather optimized routing algorithm for aircraft. And that's when I knew that this is what I need to be doing professionally because it was such an enjoyable experience. After that, I prepared for coming to the us. I packed my bags and I ended up at the school, which is most famously known for being number one in innovation at Arizona State University. And I was admitted for a master's in software engineering.
So I went through the course, it was fun, it was nice met a lot of great people and I got onto a project that Arizona State University was hosting or rather leading it was a 10 million ULI grant from nasa. And the objective of the project was to improve prognostics and reliability for air traffic management and risk assessment. And I, I went and met the professor and I got on onboard the project. I started as a research assistant. It was fun. I enjoyed. And then there was another company from Silicon Valley named Optimal Synthesis who was working on the same project. And through my professor, under whom I was a research assistant, I got to know about this company and I basically got absorbed into this company working on the same project. Since the company was already familiar with my work, it was kind of easier for them to get me on board and that helped.
So it was the biggest challenge in this entire sort of ordeal was being able to be employed as a non-citizen working in aviation because that is one of the most difficult things to do in the US because most aerospace is some in way, shape, or form related to defense. And it usually requires at least a green card or citizenship. But then I was lucky enough to find a niche where I was able to fit myself in and I worked here for the last four years and every single day has been an absolute privilege to be able to wake up and go to work. It's great. So that's my background and currently I'm also looking forward to a prospective PhD in this field sometime in the near future.


So you said that transitioning from Arizona State University and the project with NASA to the partner company who was working together on that project. So did you help a specific position in a or area of expertise as a software engineer or you generally worked on multiple research areas? How does that work out for you?


I didn't have a particular focus area. My general area of work was in data analytics, data science and machine learning, which all fall under the umbrella of artificial intelligence. But then I never focused my scope to a particular section of software engineering. I just went with every single opportunity that came to me, I took it and the one thing I learned early on is to have the ability to learn on the fly. You could know nothing about a particular opportunity that is coming to you, but then you take it and you learn on the fly while you pursue that opportunity so that you never have to miss out on anything. So to answer your question, no, there was no specialization. I just tried to do everything that came my way by trying to quickly learn.


Okay. And since you have the experience and because you worked in the ATM system, can you just explain how generally the ATM system works in the US because everybody's perception is that it's little different than anywhere else in the world. They have a huge airospace a really big number of operations. So just for instance, can you just give us a few examples of how the ATM system actually works?


The ATM system in the US works pretty well. The main reason being since there is so much commercial traffic flowing all over the country, it is extremely important to have a robust ATM system. So the first point I would like to highlight is how it works is everything should be seen from the perspective of one gate to gate flight plan. And the reason I mentioned this is if we have a flight going from point A to point B, we can think of it as an aircraft that is standing at a gate, it pushes back onto the ramp and then a taxis goes to the runway, rolls down, the runway, takes off. Then we have the initial climb of the aircraft going all the way to the point called top of ascent which is the rather, or sometimes it's also called top of climb where it's a point at which it reach reaches a cruise altitude and then it cruises and then it comes down at the other airport and then finally gets to the gate.
And we need to repeat this 10,000 times because at any given moment I mean 10,000 is all over the world, but you just repeated thousands and thousands of times and you get all the air traffic in the US because this is how generally an aircraft operates in flight. So this is the most critical component of the ATM system in the US which is the air traffic itself. Now we have the controller network in which we have the whe when it comes to terminal area, we have TRACON centers. So it's basically a concentric circle format of how these aircraft are routed and helped with navigation. That's the second aspect of the ATM system. Third aspect I would like to point out is all the safety protocols in place, be it TCAS, which is the collision avoid system be it the terrain avoidance system all the other alert systems, some of which come from aircraft manufacturer for the equipment safety.
So safety is a hugely critical part of how the A T M is carried out every single day because any given moment, especially at busy airports, you take New York, jfk, you take Atlanta. Atlanta, which I believe is the I think it's called Heartfield Jackson Airport. It's the most, it's I guess the busiest airport in the world. You have San Francisco airports like these, they literally have a takeoff or landing pretty much every second of the day or rather every minute of the day. So with this kind of a system air traffic is managed all over the us and then unfortunately, of course incidents happen, sometimes even accidents happen, but then things are learned from those incidents and accidents and changes are made to the system to not repeat that again. So we can think of it as a feedback network. Something happened, the output of that is given back as input to the system to make some changes to improve the system a little bit more than it was safer before the incident. So that's how the ATM system in general works in the United States.


So it's interesting to hear how the FAA, it's managing the air traffic. You mentioned the complexity of operation, the number of operation, everything that basically falls under the everyday gate to gate controlling. But just to go back since you mentioned that you got into this industry from a NASA project. So an interesting question would be how FAA and NASA work together in the air traffic control system within the us. So it's managing the traffic, they're responsible for the safety, all the controllers are coming from the FAA, but what's the role of NASA? We can read all this titles that they dedicated certain amount of money for research and development, but since you were there and saw it in from your perspective, can you just tell us to the listeners on this podcast, what's the real difference between NASA and FAA and how NASA contributes in the overall ATC system in us?


Yeah, so the faa, which is the Federal Aviation Administration, as the name suggests it's main job, is to look after commercial aviation in the country, A commercial among other aviation also. And NASA is primarily a space agency that deals with stuff that is an outer space. However, there are points at which the operations of the FAA and NASA overlapped because a lot of things if you think about it actually came out of the space program out of nasa. For example wireless power tools were used in space and now they're used on earth. Grooving on the interstate. Actually if you go on a interstate freeway and if the turn is a bit too steep, the road is going to be grooved. That idea was taken from NASA because when the space shuttle landed after going either to the international space station or to the Hubble telescope and it landed, they found that the because shuttle doesn't have engines, it's basically a glider when it comes down.
When it landed, they found that it was drifting across the runway. So they decided to groove the runway and the tires would stick in and it won't drift away. And that idea was taken from that NASA program and apply it to freeways that we all used today. Similarly, there are things from the space agency that could be of help to civil aviation or commercial aviation in this country with the FAA. So the key difference between what FAA does and NASA does is FAA runs day-to-day business, make sure aviation runs in the US and NASA can be thought of as the brains in certain aspects when it comes to safety. Since NASA already has all this expertise, experience and data from its space program that is being leveraged by the FAA to improve commercial aviation performance and safety. So this particular contract was in an initiative by NASA under the U L I, which is called University Leadership Initiative and Arizona State University won this 10 million contract and that's how it works. So it all goes hand in hand because there were people from FAA and at least a dozen, not even a dozen, actually 20 to 25 other premier aerospace organizations and universities from the US that were a part of this project. So that's how NASA and FAA work kind of overlap through mutual collaboration.


So it is great to hear how NASA and FAA are back in the front of an ATC system and to understand how specific ATC system is being developed. But when we talk about developing a new system or conducting research within this area, you mentioned yourself that there are around 20 something organizations that contribute to this initiative, but besides the initiative itself is working with NASA and FAA, is that allowed, for example, a smaller company that works in aviation to join the initiative or to join on the journey work, working on a specific project regardless if it's a safety net project optimization project? Any number of topics in A T C for example,


If I understand right, your question is, is it allowed for a small scale company to compete to work with a gigantic organization like nasa?


Yeah, exactly. And what is the process of actually getting in touch start to work for the ATC system?


Oh, okay. Yeah, it's absolutely allowed. As long as it's a registered company and they've gone through all the protocols and they have all the clearances required, then yeah, if they apply for a contract and they win it, they're totally allowed. There might be a certain threshold of the minimum number of employees for a company, but as long as that is satisfied, they're allowed to do the work. And second part you said is the process about how this is done and how that works is generally these projects are put out as contracts and a lot of companies compete to win this contract. Sometimes one company wins the entire contract, sometimes two companies win it. So it is split 50% for each and the companies have to write proposals to win these projects. So the companies can show how they have done similar projects in the past how they have the resources to accomplish this project skill sets of the employees of the company who will whose expertise will help take this contract to a logical end. Things like these are mainly these contracting agencies like nasa, FAA, even the other airospace agencies they look for in a company to be able to do this. Now in some cases, one may argue that huge corporations like Lockheed Martin, Boeing, they might be having and upper hand when it comes to contractors like this just because of their scale of operation. While that may be true in some cases, but still even small scale companies do stand a fair chance when it comes to competing to win one of these contracts. It's definitely doable.


And since we are talking about this, and you mentioned some specific security restrictions might apply in you yourself being an international student getting a job in a company and in the accessing this field how hard was this process? So I guess you were lucky, but can you just explain a little more on the journey with all these security restrictions and international company or student coming in the US and look for job in this field?


To put it in simple words, it is extremely difficult to find a job in aerospace. If you are an international citizen, if you have a green card or citizenship, you're pretty much set. But then if you have anything less than a green card, I refer to it as being able to walk on water. That's how difficult it is. But then the reason I say this is because throughout my master's degree I constantly was applying to companies like Honeywell, Boeing, you name it any single aerospace company I found I applied to it, even small scale ones, even startups. And literally nobody would accept my even look at my application if I say that I need visa sponsorship because I'm an international candidate. That is the biggest problem from a candidate's perspective if they want to work in US aerospace. But however though, not to discourage anyone who is planning to enter this field but if you try and keep trying, just grow the network that you have, know more people who are working in this industry and develop skills that are crucial and critical to make you stand out of a crowd sets that only you are the one in the visible pool of candidates who can get this job done.
And that goes a long way. In my case, since you mentioned how it worked out is since I started building a software with my professor under whom I was a research assistant at Arizona State, this company saw my work over the last one and a half years and how successful our work was. So at that time, my potential employer, potential future employer, didn't have to really verify what I could do because he saw my work firsthand. He already had proof of it. So that's the way to go about it. If someone tries to just apply to an aerospace company and expect that they'll get back big, I mean international student, that is generally not a very good use of the time because you can use that time for something else, probably. But then it is difficult to answer your question. However, with extreme determination and hard work, it is possible.


And thank you Hari for that. It's really inspirational for some of our younger listeners to hear how a person from the studies then continues in the aviation industry, specifically in air traffic management. To follow up on your education background, as a software engineer, you went to school for this, but how did you manage to be professional in aviation and knowledgeable in the air traffic management field specifically? Did you take any classes? Did you like self-taught yourself? How was that process?


Right. So I did not take any classes professionally or as part of my curriculum. It was all self-taught just out of sheer interest because the number one best resource on this planet to learn anything is the internet, particularly YouTube. YouTube contains so many different channels from universities like MIT, OpenCourseWare, you have so many really, really talented, skilled and qualified people trying to teach you complex and complicated concepts. So that's where that was a starting point for me. And then eventually I went on to also read books, reading lot of books, because the truth is what really matters. I mean, of course when it comes to licensing, you need a degree you need professional and an official credential, but then when it comes to knowledge, you can gain it from the internet. You can literally gain unlimited knowledge from the internet. And that's what helped me because ultimately in the interview, more than the degree, they usually tend to look for what you can do and what you have proven ability of.


And just to give us, to give our listeners a better idea, so you worked for this smaller company that was partnering in the NASA initiative for ATC research. What kind of projects you worked on? Just very brief summary of the kind of projects you worked on, if you can name a few.


Sure. Primarily with this company I worked on this NASA ULI project, which was to improve prognostics and safety when it comes to assessing reliability and operating of the US national airspace. That was my primary project. Apart from that I did a few more projects with a couple more government agencies. Unfortunately because of contract terms and the security involved, I'm not able to name those agencies. I did some hardware projects with them that involved embedded systems working with sensors inertial navigation systems, which are basically used for vehicle guidance. So that was my other project that I worked on. What else? Oh yeah, I also worked on a genetic search toolbox for matlab, but that association was fairly brief. That was a bridge for me between two major projects. And then ultimately we ended up open sourcing the NASA project that we built. So my company, optimal Synthesis Inc, built a flight simulation software called Gnats.
It stands for generalized National Airspace Trajectory Prediction System, which is now open sourced on GitHub that anyone can just go and look it up. It's called G N A T S. And the reason we open source it is because since our customer, NASA was satisfied with our project, we decided to expand the scope of reach of the software. The most critical part was handling the data that was confidential only for nasa. So we created a scrambled open version of the data so that anybody can use it, anyone can insert their own data and play with the software. So that was a whole another project towards the end of the fourth year at this company so far. So those are the major projects that I worked on at this company.


And were this project so since it's a leadership initiative so there was a research component as well and there was a real development that it is used down the line, correct?


That is correct, yes. There was a research and development parallelly side by side.


Okay. Since me and you, we go way back from Arizona State University and know each other, working for the same professor and since at the very beginning you mentioned that you are transitioning now from work environment, going back and doing your PhD. I know that you will be doing your PhD in human factors. So can you explain us a little bit about the journey, like going from this research that you did and going to human factors and back to PhD or Academy environment? Again, why and how did you decide to make these changes?


Sure. So one of the critical aspects of this project, of the NASA project that I worked on was to model human error, to model pilot error, air traffic, controller error, and ground operator error. Because no matter how sophisticated a system is, as long as the human interaction is not smooth, everything pretty much goes out of the window. For example, we can build the most sophisticated collision avoidance system, but then if a pilot does not turn it on, for example, or the collision avoidance systems usually is not a great example honestly because it usually is automatically turned on. But then some safety features need to be turned on by the human. What if they're preoccupied with something in their head and they forget to turn it on? The system essentially is redundant from that point on. So we have so many things that could go wrong when it comes to a human brain.
What technology and how can we make the human just a little more cognitively sound and present in the moment of operation to improve safety in the field of aviation? So you mentioned human factors. Yeah, that is the broad spectrum of the program that I'm joining it. It's official name is human systems engineering. But then my focus area is going to be what I just mentioned to make sure that the aviation human systems can be enhanced just a little more if possible, to get a deeper understanding of what the uncertainties are when it comes to the human brain. So the reason I'm transitioning back to a student visa and trying to go into PhD again is because over the last four years, the amount of fun I've had with this job, the passion, the rage and passion I feel for working every day has made me realize this is the field I want to be for the rest of my life.
This journey started when I was four years old. In fact, if I might just mention the anecdote, I remember I was four years old and I had gone out with my mother somewhere and we were walking on the street and we used to live very close to the airport because my grandfather, he was the director of the Indian meteorological department. So our house was very close to the airport. And I remember walking back to the house, I looked up at the sky. This aircraft came into land since I was four, I don't know what kind of an aircraft, it came into land. It was just above our heads because we lived very close to the airport and it beautifully deployed its landing gear and it came in and landed. I remember I was four. I looked at that aircraft land and at that moment I told myself, whatever makes this happen is what I want to work in.
I'll never forget that moment. And last four years literally just cemented that thought in my head that this is what I need to be doing for the rest of my life. And getting a PhD would not only expose me to more knowledge, but also more people. Because as you mentioned, we both went to Arizona State, great school, great people, great professors, so we already know the professors. And also these are the same professors who I've been working with over the last four years in the course of this project since they're from ASU, who is now our customer. So that's where this entire idea came out of. And also a side benefit of this whole exercise is if you get a PhD as an international student, your green card process gets accelerated pretty quickly. So that was the main reason, trust me. But then that is something that is definitely a side benefit because the quicker you get a green card, the more you can advance your career in aerospace. So I definitely look forward to it as I'm waiting for my student visa to come sometime. So that's what my goal was with the PhD.


Okay. So thank you for all this passion about air traffic control and how everything works. And it's a great insight for the students, students who will eventually listen to this podcast just to get a little bit of more understanding what it takes to be international student working in aviation in us. And just to go back on the things that you're going to work for your PhD, do you have in mind or does it exist, any kind of statistics? How many systems need or what is the rate in percentage expressing in percentage of mistakes that are done because of the human and that cause really big safety concerns, let's say?


That's a great question. I would estimate it to be anywhere between 30 to 35% because if you go to the N T S B website, you can get a lot of these accident reports. These are or rather the investigation reports of aviation accidents. And when we worked as research assistants with our professor at ASU, there was a mini project that was done. I led the project in which we took all these reports and we did linguistic analysis to find out, well what really caused each of these accidents. It was a machine learning project and the number came out to be around 30% where pilot or the air traffic controller was the one at fault for the accident. This can be an action lag wherein the pilot should have done something at a certain point but waited for a good amount of time before executing the action.
That's one example. Controller clearance delay wherein the controller should have asked the pilot to go around, but instead it was delayed to a point where the pilot could not go around and they had to come into land. And that created a safety hazard and potentially an incident after that you have erroneous parameters. For example, the controller could say name an airline a airline 1, 2, 3, turn left heading 0 4 0. And because the air traffic controller's mind is preoccupied with something else, instead of 0 4 0, the pilot is asked to turn 0 5 0. That turn itself, even though it's only 10 degree difference in over time, it could magnify and potentially come in the path of another aircraft. So all these are risk factors caused by human either carelessness or error. So what technology and what measures could prevent this is what we are trying to look for. And the main reason is because the human brain is the biggest source, is the most prominent source of uncertainty. If you program a computer to do something, it'll do it as long as there are no mechanical or electrical failures. But then the human brain literally has infinite possibilities to operate in a certain arena, in finite possibility of ways in which it can operate in a certain arena. So we are trying to pin that, we trying to nail that down so that we make the human behavior more predictable so that we can improve the prognostics for aviation safety.


So it's a great thing to hear how machine learning human factors and everything can contribute to avoiding the human errors in the air traffic control context and many accidents to be avoided, some of them and even that are fatal to be completely avoided. Since you have all this experience, and since we are at the very end of our podcast today, with all of your experience, can you tell us and this is our traditional question, according to your opinion and to you, how do you think that the A T C technology world environment will be evolving in the next five to 10 years and in the next 50 years from now?


Yeah, that's a good question. So the number one biggest change I think we'll all see is increased research because more and more people are getting on board. That's number one. Number two is I think we are going to see more automation because whether we like it or not, automation in a lot of cases actually reduces uncertainty because uncertainty is the worst enemy of aviation. Not just like stock markets. Not knowing what's going to come is worse than knowing that something bad's going to come because if there is something bad that is impending some risk that is impending, we at least know and we can take measures. But not knowing what's next is the worst. So I think we gotta have a lot of automation. Secondly, thirdly, I think we we'll have simplification of a lot of processes. To give you a couple examples right now in the US as I'm sure even in Canada might be the case where we have named waypoints, like DAMBA or BOILL, these are all waypoints in California, they're being replaced with G P S coordinates.
So latitude coma longitude and optimization that is being planned is only give the last two digits of the latitude and longitude. Because if you're in the San Francisco area, for example, the first five digits of your latitude and longitude remain the same. It only changes in the last two digits. So simplification when it comes to operation is going to be a critical factor. And to also go back to the automation point: A D S B, which is called automatic dependent surveillance broadcast, which is the information that the aircraft keeps sending about its latitude, longitude, altitude, speed, mean air speed, vertical speed and course is constantly being beamed from the aircraft to ground. You can buy dongles on Amazon when you can retrieve flight data, which is something we did at our company not too long ago that was made mandatory by the us I mean by FAA, I believe on January 20th, 2020.
I think that was the date when that happened. So that was made mandatory for all airlines to follow, including regional jets. That all aids in advancing the course of automation and data simplification. And there's always an increasing trend in how many people need aviation to function smoothly. Passenger traffic is growing exponentially. I mean it has grown exponentially over the past couple decades at least. In fact, in January I did a presentation with I A T Bombay in San Diego where we highlighted all this data about passenger traffic. Of course, due to the coronavirus, it has dipped down, but it's going to come back up before we know it. So over the next five to 10 year mark we need to make sure that the A T M industry in general needs to make sure that the customer demands are met because it's a huge business risk that not risk, but a huge business endeavor that that's being run here.
And I think a lot of universities are going to come on board because Arizona State University, and I need to give props to the president Dr. Michael Crow, a fantastic president of one of the best university in the world, as I'm sure Stojanche, you would agree, number one in innovation for the last seven years. I think because so many endeavors are being pursued by ASU that so many contracts, so many projects from all across the spectrum, all across the industries they're coming to ASU and he's leading ASU the right way as such as so many other universities who are getting on board with such thing. In fact, I think Northern and Northern Arizona University now is also going on board with projects like this. So I think we are going to have an increased involvement of students and researchers to prepare the next generation workforce to advance the course of safer, more efficient, and a more scalable ATM system.


What do you think for the next 50 years from now, what would be happening to the ATC environment, the technology and so on?


Right. 50 years is a long time when it comes to the scope of expansion, when of air traffic management in general. So over 50 years, what would happen is, first of all the things that we discussed about over the next five to 10 years, they'll just keep growing for the next 50 years. But I think there's going to be a system in place wherein it's going to be much more easier to do business as a commercial jet liner, but also regional jetliners who today sometimes have a hard time conducting operations because of the immense competition that they face. So I think over the next 50 years, one of the things that could happen is the playing field becomes a little more leveled for these smaller businesses too, to have an equal cut at the opportunities that are available. I think that's one thing that could happen over the next 50 years because that's a little bit kind of a medium term goal for them.
Secondly, aviation safety itself would be more sophisticated when it comes to the other aspects of what we are trying to preserve. Because today, if there is an air crash or an air incident, the number one and the most important question that is being asked how many people are injured or how many casualties occurred. But I think over the next five to 10 years when human loss and human safety is put on the forefront would've reached a point wherein now we can focus on the other things that we need to say, for example, cargo. If there have been very famous accidents of cargo airplanes and people generally, I'm not saying this for everyone, but generally it's a perception that a cargo airplane loss is not as bad as a passenger aircraft that was involved in an accident because cargo, as long as the crew is safe, everything else is just cargo, not human beings.
But then I think we would reach a point where an even cargo loss or even collateral loss, which is a better term, would be seen as big as a threat as human loss because it's how the efficiency would advance over the next few decades or 50 years, as you said. And the third thing is I think an upgradation of technology itself, because even now a lot of the technology that is being used for avionics, I mean things that are put on an airplane are well tested and they are sophisticated. But then when it comes to computational power, I think technology is going to be massively upgraded. Lot of, especially when it comes to government, lot of stuff is from the eighties when it comes to technology. So there's going to be a, I think the entire ATM and aerospace industry is due for a technological overhaul wherein we increase computational power manyfold. I think that's going to happen within the next 50 year mark. So these are the major changes that I see happening in aviation in general over the next 50 years.


Well, thank you Hari, for the answer of the second part of the question.


Sure.


And as I said, this would be our last question for the podcast today. Thank you very much for your attendance and we really appreciate that you accepted our invitation to be our guest today.

 

It, it's been a real honor a pleasure. We met at ASU. It's, it's, it's good talking after a while, and I really appreciate you guys for having me on. Thank you.