#56 Robert Rose, Reliable Robotics: Exploring the future of autonomy in aviation

Welcome back to The Vertical Space and our conversation with Robert Rose, co-founder and CEO of Reliable Robotics.  

This is a conversation about autonomy. We discussed the arc of automation and, what evolved in the conversation, the arc of safety in aviation from its early days to today. You'll notice how Robert will closely and consistently link autonomy to safety in the beginning, throughout and at the end of the conversation; an irrefutable argument.  He essentially says autonomy is going to happen so get used to it. But he makes the hard medicine taste just a bit better by tying the key arguments to safety. 

We spend a bit of time on the motivation for pursuing autonomous flight by segment; GA, Part 135/121 with a safety imperative for automation being more required in the former than in the latter. After a few attempts, we eventually get to other reasons why autonomy is needed in addition to, and beyond the safety case. As we've discussed many times the podcasts, we discussed the need for autonomy for many of the advanced air mobility business models to scale, and to meet their financial projections.

Robert discusses the autonomous capabilities of Reliable Robotics, and as Robert says, what makes them unique is their ground up designs for automation and certification, where it's more difficult for others to go back and design for automation if that wasn't the original intention. We spent a lot of time discussing the processes required to certify automation systems, and get into a really interesting discussion around the design of their air to air radar, it's value, market size, and why build it on their own versus purchasing off the shelf radar and whether or not it's core to their autonomy focus or perhaps a distraction to that focus.  He wraps up his talk with great advice to entrepreneurs. 

Chapters

• 6:13 Robert's Opinions on Autonomy
• 11:11 History of Automation in Aviation
• 18:10 The Motivation for Pursuing Autonomy in Different Aviation Segments
• 23:43 The Safety Case for Automation
• 27:47 How Reliable Robotics is Different Than Current Remotely Operated Military UAS
• 36:39 Convincing the Regulators
• 43:51 Certification Case Study: The Autoland System
• 47:00 Software Behind Autonomy and Addressing Unforeseen Situations
• 49:20 Certification Process and Avoiding FAA Denials
• 51:48 Market Potential and Business Model for Autonomous Aircraft
• 1:01:13 Developing In-house Radar Capability
• 1:10:23 The Role of Primary Surveillance Radar in Airspace Integration

Transcript

Robert: 0:27

Collecting data is a trigger phrase for me. You triggered me. It really irks me when people use that expression because I think, well, for what purpose? Why are you collecting data? What system is this that you have where enough data on its functioning makes a safety argument? I'm not comfortable with a system where people say, well because it worked for several thousand hours or because we did so many auto landings with it, it's safe. That's just bogus. That's not the way these processes work. That's how you get to 10 to the minus two, maybe.

Jim: 1:04

Hey everyone. Welcome back to The Vertical Space and our conversation with Robert Rose co-founder and CEO of Reliable Robotics. Listen, few have Robert's pedigree to create systems for autonomous aircraft with his work and leadership at Reliable Robotics, as well as from his position as Director of Flight Software at Space X and at Tesla, where they worked on bringing to market the first consumer automobile with fully unassisted self-driving capability. This is a conversation about autonomy. We discussed the arc of automation and what evolved in the conversation, the arc of safety in aviation from its early days to today. You'll notice how Robert will closely and consistently link autonomy to safety in the beginning, throughout and at the end of the conversation; an irrefutable argument. He essentially says autonomy is going to happen so get used to it. But he makes the hard medicine taste just a bit better by tying the key arguments to safety. We spend a bit of time on the motivation for pursuing autonomous flight by segment GA part 1 35, 1 21 with a safety imperative with automation being more required in the former than in the latter. After a few attempts, we eventually get to other reasons why autonomy is needed in addition to, and beyond the safety case. As we've discussed many times the podcasts, we discussed the need for autonomy for many of the advanced air mobility business models to scale, and to meet their financial projections. Robert discusses the autonomous capabilities of Reliable Robotics, and as Robert says, what makes them unique is their ground up designs for automation and certification, where it's more difficult for others to go back and design for automation if that wasn't the original intention. We spent a lot of time discussing the processes required to certify automation systems. And you'll notice that we make every attempt to quantify the addressable market for autonomy and for Reliable Robotics at each of its stages today and in the future. Robert deftly steers the path to the size of the overall aviation market. We try to stay focused on the addressable market in the next five years and what opportunities within Reliable Robotics are tied exclusively to the Caravan and how easy it is to transition the capabilities to other aircraft. Next is a really interesting discussion around the design of their air to air radar it's value, market size, and why build it on their own versus purchasing off the shelf radar and whether or not it's core to their autonomy focus or perhaps a distraction to that focus. This naturally leads to a discussion around their airspace integration solution, including capabilities from their own radar and integrating data from other sources and the dependency on those solutions. And a discussion around the availability and value of primary radar data, where some would say, including me, it's availability is a lot further away than most people think especially given justifiably how long it took to release secondary radar data to industry. Listen to what Robert says is his biggest upcoming challenges if he had to pick one. He wraps up his talk with great advice to entrepreneurs. Robert, thanks for joining us and to our guests, we hope you enjoy our talk with Robert Rose as you innovate in The Vertical Space. Robert's deep engineering experience spans over 20 years across aerospace, self-driving cars and robotics. Prior to co-founding Reliable Robotics, in his role as Director of Flight Software at SpaceX, Robert was responsible for flight, ground, simulation and data management software successfully launching and operating the first 10 Falcon 9 rockets, 5 Dragon spacecraft, and Grasshopper VT VL tests. He led the development of the onboards flight software for their first commercial mission to the international space station, for which he received special recognition from NASA. At Tesla, Robert brought to market the first consumer automobile with fully unassisted self-driving capability. He led the team through developing novel computer vision, human machine interfaces and control technologies resulting in the successful deployment of the autopilot 1.0 and the 7.0 version of Tesla's instrument panel. At Google X, Robert led a development team on a secret project combining AI techniques with practical robotic systems to bring advanced machine perception and manipulation technologies to large vehicles. Earlier in his career, Robert contributed to three"Game of the Year" award winning titles during his time at Sony PlayStation and worked at HP as a software engineer. Robert is a presidentially invited Fellow of the Royal Aeronautical Society. Robert holds multiple patents, a BS in Computer Science, a BS in Computer Engineering and an MS in Electrical and Computer Engineering from Oregon State University. Robert, welcome to The Vertical Space.

Robert: 6:10

Hey, great to be here. Thank you for having me.

Jim: 6:13

First question, is there anything that very few in the industry agree with you on?

Robert: 6:17

So there's a couple of different ways I could take this. I could take this, the, are you asking, Robert the individual or Robert the company? let me take the, company perspective. we started Reliable Robotics because we felt that somebody needed to cut a different path from everybody else in, in the UAM, AAM, RAM, new aviation space. We felt very strongly that somebody needed to focus specifically on autonomy and autonomy in the United States, because at the time, and I think this is still true today, I still hear from folks, people say, well, well, this is never going to happen. We're never going to have autonomous aircraft or this is never going to happen in the United States. It's going to happen in Dubai or New Zealand or Brazil or someplace like that first. Or people say, well, nobody, even if we have autonomy, nobody's going to buy it. Customers aren't going to want it. Pilots will never allow it. Passengers will never accept it. And I feel that a lot of these criticisms are still true to some extent today, or at least many people still hold these, and bring up these criticisms. We obviously, I obviously feel quite different. I believe this will happen. It will happen very soon. It will happen in the United States. I believe there is a path to bring this technology forward that, pilots and other stakeholders in aviation will accept. And we can talk more about that later. And I think ultimately we'll get to a place where passengers are going to accept it. this is the future. I think the near future is going to have orders of magnitude more aircraft in the sky, and the way that we get there is with greater levels of autonomy.

Jim: 7:50

when you say it's here, it's going to happen. and people are going to accept it. So what level of automation are you describing when you say it?

Robert: 8:00

I'm referring to automated aircraft that enable you to take the pilot out of the plane and into a control center, place them into a control center. This requires an aircraft that's capable of fully automated taxi, takeoff, and landing.

Jim: 8:14

Now you mentioned, you have, you had a professional response to those who, don't always agree with you on something. And then you said you had a personal as well. So what's the personal?

Robert: 8:24

Well, I want aviation to be safer. I'm a pilot and I think there's a lot of people that get into aviation because they have that love of, grabbing the controls and, they are the airplane, be the airplane. And there's a part of me that I get that, I understand the appeal. I think, unlike many that get into aviation, I also appreciate it for what it enables you to do and where it enables you to go. And I value the where you can go with aviation more than the experience of aviation, if that makes any sense. And for me, I want more automation in the cockpit. to ensure my safety, the safety of my family and others that I take in the plane with me. I've had several occasions in my limited experience flying where I've gotten myself into trouble and I wish there was a land now button in the aircraft. I think there should be a land now button in all airplanes. not just a system that's certified for emergency use only, but a normal use Autoland gets you back into a safe configuration and on the ground. and we have aspects of this today in airplanes. we've got wings levelers and emergency autoland and autopilots that can do heading changes and capture an approach and do all of that, but I don't think that goes far enough. I think there's a lot more that we can do to make aircraft safer.

Jim: 9:51

Robert, before we get into, the meat of the discussion, you used the word safety now, 15 times in a couple of minutes,

Robert: 10:00

Oh,

Jim: 10:00

two or three, exactly two or three

Robert: 10:02

Thank you for noticing.

Jim: 10:03

two or three years ago, would safety have been mentioned as much as the use case for automation or what was your value proposition a couple of years ago for automation and fully autonomous flight?

Robert: 10:16

I feel like we've been consistent on this from the beginning. Now, I think the larger conversation this industry has been having has been shifting. And I think that's because folks or others are recognizing that, the way you get the FAA and larger customers excited about this technology is you do that through safety by improving safety. I don't know if that answers your question. is that what you're getting at?

Jim: 10:41

Well, there's a piece of me that says, well, you can't argue with safety. although I question whether people will pay for safety, you can't argue with safety. But yes, the many people listening right now will think, oh, we're going to talk about automation, we're going to talk about fully autonomous flight, we're going to talk about the economic value, we're going to talk about pilotless flights, which is extraordinarily controversial. But in some ways, you've steered a path here that is not controversial. Who can argue with safety? And you're saying that's a deliberate act for a variety of reasons.

Robert: 11:11

Yeah, if you look at what stakeholders in the aviation industry care about, safety is the first thing that they talk about. If you look at, well just step back and look at the history of technology advancements we've had in airplanes over the last a hundred plus years. why has anything in an aircraft changed? It's changed to make the airplane safer. Like that's not an accident. I actually would be interested in knowing about improvements that were made strictly for economic reasons. Okay, maybe winglets come to mind. But most technologies, or at least electronic, avionic technologies, are there to prevent accidents and save lives. Period. And I think autonomy fits within that same trajectory. Autonomy or autonomous aircraft. It is just, it is an incremental set of enhancements on top of the incremental enhancements we've been making over the last several decades. I don't see the advanced navigation system that we're developing a whole lot different than what has been done in previous years with GPS, ILS, and other radio navigation systems, and even going all the way back to light beacon navigation systems. these are all technology enhancements to help improve safety. the automated landing system that we developed, I think, fits within that same trajectory of, automated landing systems and automated approach systems that have been developed to date. Our automated take off system, which is really a take off rejection system, fits largely within what's been happening in the industry for, for decades on automated monitoring of aircraft configuration during take off and take off rejection systems, And finally, automated taxiing, I see that, as a bit more new and novel, but from a core technology standpoint, really just builds on systems that are already going into aircraft. and even now, it's, our automated taxi system, I would say, builds on taxi planning features that you have in an EFB now, and it's just taking that one tiny step further, to ultimately improve the safety of you, the pilot, the occupants of the aircraft, and the people that operate around that aircraft.

Peter: 13:25

I'm sure a lot in our audience all have, pretty good familiarity with aviation history and the milestones that were achieved over the last century. But from your perspective and from the perspective of what you're building at Reliable, What are the key milestones along that arc in terms of putting in place the building blocks and setting the precedence for autonomous flight?

Robert: 13:51

yeah, we, and you're referring here specifically to the history of aviation. what are key milestones throughout the last century?

Peter: 13:58

Yes.

Robert: 13:58

Okay, Full Authority Digital Engine Control. Huge. I think we need FADECs in every plane. I think it's ridiculous that small aircraft, expect you to sit there and manage mixture and, prop speed and all of those other components. it, it should just be, fully automatic. At the very least, we should have auto throttles in every plane. Instrument landing system goes back to the 40s. I believe the first experiments were in the 40s with automated landing systems, and then that technology was further developed in the 50s, and then commercialized first in the 60s. I think it was the, I think it was the Trident, I think it was British Airways, I could be wrong, that first deployed that technology, and that drove, developments in fault tolerant computing, I think, is something we take for granted because the space race was occurring around the same time, but the first real commercialization of a fault tolerant computing system was an Autoland system. This goes back to the 60s. NASA in the 70s built, I can't recall the acronym, I just remember that it wasn't the best choice for an acronym, but they built a, oh, it was the Terminally Configured Aircraft Program. they built a flight management system that allowed you to specify a flight path through all phases of flight, takeoff, enroute, and all the way to landing. And they actually flew this, it was on a Boeing aircraft. I can't recall exactly which model. GPS also stands out as another huge safety benefit. switching from, ILS approaches, which require expensive, infrastructure to be installed at the airport and then maintained. By the way, the FAA has a small army of pilots that just do nothing but fly around the country to monitor approaches and to make sure all the radio signals are correct. Moving away from that and towards a satellite constellation system has gone a long way to improve safety because it's enabled so many more airports to have RNAV approaches. That's huge! Think about all the accidents that we had before from people messing up approaches and going visual before they should have, or scud running. a lot of that has gone away. It's gone a huge way to improve safety.

Jim: 16:04

let's stay on the history of flight. Talk about the arc of automation and aviation from the early days.

Robert: 16:10

Well, we go all the way back to the Sperry autopilot, would be the start of it. I believe those tests were done first in the 30s, and then instrument landing system, as I mentioned, fault tolerant computing associated with that developed in the 60s. I think other key moments, key inflection points would be, Automated descent systems that monitor depressurization events, and then will bring an aircraft back down, to a safe altitude. I don't want to get too dark, but there have been some accidents recently where if that technology had been in those Private aircraft, it would have saved lives, but it does exist in a few large, commercial aircraft. there's a GA Avionics manufacturer that, promotes this as a product. I don't know off the top of my head how many aircraft have actually adopted this system, but, That, I think, is huge. for us, that's very significant because, automated descent following a depressurization event, you've got there a system that at very high integrity is monitoring for an adverse condition or an anomalous condition in the aircraft. And without asking the pilot, it's taking over the plane and putting the plane into a safe configuration. It's descending down to a safe altitude to save the lives of everybody that's on board the aircraft. And we talk about, that as like a key inflection point internally, because that's a time when you've got, an automated system taking over from a human, and in a split second manner, acting to take control of the aircraft. and you can't find a straight line between that technology and the regulations. the closest regulation being 91. 3, the pilot is the ultimate authority for the safe operation of the aircraft. here you've got the automation taking that responsibility, and that's significant precedence for us. and a lot of the system that we're developing now leverages on that precedence, or builds upon that precedence, I should say.

Luka: 18:10

Robert, as we pursue, autonomous flight. Let's talk about the motivation for, this goal and let's do it on a segment by segment basis. So describe for each of the, part 121 operations or 135 or general aviation or UAS systems for that matter. Why do each of those segments really care about fully autonomous flight?

Robert: 18:34

Okay, so GA, 121, 135, and then you had UAS, which I assume you're referring to small UAS

Luka: 18:43

correct.

Robert: 18:44

okay, sub 55 pounds. The Box Droppers, as we affectionately call them. GA, I think, has a clear need to improve safety. there's this myth that is commonly propagated that aviation is the safest mode of transportation. And that's true for certain 121 operations. That's definitely not true for GA. And there's a couple of papers that were published in the last several years that dug into the data here and concluded that, flying small aircraft, and this is GA, and certain 135 operations is actually 8 to 15 times more dangerous than driving on a per mile basis. I found this shocking. my intuition was planes are probably not safer than driving smaller planes. I didn't know it was this much worse. and so there, the motivation is to improve safety. That then ties to a very clear economic or business imperative. if you talk to almost any small airplane manufacturer, they will tell you that the biggest impediment to them selling more aircraft is the significant other. I don't want you getting in that thing, or I'm not getting in that thing with you. And systems like, parachutes and whatnot have gone a long way to alleviate some of those concerns, but not all of them. My strong belief is higher levels of automation are going to improve safety and thus expand general aviation to even more people, which I think is very exciting. I know my significant other, my wife, would be very excited about this and much more comfortable flying with me if she knew that we had a land now button.

Luka: 20:22

I agree. General aviation does have a much more pronounced safety problem, but what about part 121 operations and 135?

Robert: 20:30

I'll take 135 next. I'll take these in, in order from smallest to biggest. So in 135 operations, you also have a safety imperative. Here you've got, aircraft that are frequently operated single pilot, frequently operated at lower altitudes in, More adverse weather conditions than what you would see in 121 operations. And the nature of the work, tends to be a little bit less, What's the word I'm looking for? A little bit less structured than what in 1 2 1 operations. safety management systems, for example, are not required of all 1 3 5 operations. They should be, but they're not. small 135 operations don't have the FAA requirement. certain operators impose. They voluntarily adopt SMS. I think that's great. more operations should do that. But there, I don't want to get dark again and start citing specific accidents, but if you Google around, 135 accidents, especially 135 cargo accidents, you'll see some very unfortunate events where, People have gotten themselves into some pretty sticky situations that could have been prevented with automation. so there's a safety imperative there. The business, motivation in 135 is a bit different. I'm going to say pilot shortage. that shouldn't be shocking to anybody, but the pilot shortage in the 121 realm is trickling down to the 135 realm, and it is vacuuming up pilots out of 135 operations at a very fast rate. I hear from operators quite frequently that turnover is extremely high. it's getting harder to recruit and retain people. Salaries are skyrocketing. And on top of all of this, because of the accidents we were talking about earlier, insurance and training and other factors are all becoming more expensive, so it's becoming harder to find people to run these operations. For some operators that I've spoken with, it's existential. They say that they may not be able to run their operation in the near future because of pilots, and so they need to look at automation. they need to look at ways they can, Get more people, trained up, which you can do with improved levels of automation, or move to remote piloting as our technology is moving. so finally, in the 121 realm, you've got some different dynamics. pilot shortage is definitely a topic of conversation, but, I think safety, it's not talked about that much publicly, but Congress has been bringing it up as of late. we've had a few incidents that have occurred at major U. S. airports that made headlines that could have been prevented with certified alerting and guidance and higher levels of automation. And I think the industry needs that to be able to keep operations efficiently flowing and to maintain public confidence in the aviation system. I don't think you can draw a straight line between pilot shortage and automation. I'm not ready to go there. I think some folks in the industry are trying to draw a straight line, but I think it's going to be several more years before, the pilot shortage is a direct driver of automation.

Peter: 23:43

can you go a layer deeper, either in the Part 135 safety cases that you reference or Part 121, and give the audience some examples of a dangerous situation that occurred and the cause of it and how automation would have prevented it because obviously in aviation there are many different causes that lead to an accident. Some of them can be weather related, some of them can be, a failure of some part of the aircraft, or it can be a failure of communications. help us connect the dots between how automation specifically would alleviate certain types of accidents.

Robert: 24:28

yeah, I'll have to go back further because I I don't like to talk about recent accidents until the NTSB has completed their reports. I don't think it's appropriate to draw conclusions. but there was an accident about, a year ago involving, a Caravan pilot where, this hits pretty close to home for us, where they were attempting a, a landing, at an airport in northwestern United States, and the weather was pretty close to minimums, and they came down, broke out, and, were not able to make contact with, the runway, or they didn't break out, they weren't able to make contact with the runway, visual contact, they performed a go around, and it appears from the preliminary reports that I've read that, They decided then to descend, believing they understood the airport environment, broke out, and then, ran into some fixed infrastructure on the ground. And, this breaks my heart because, there's just no reason they needed to die. If they'd had a system that enabled, Zero Visibility Autoland, if they had a high precision navigation system that gave them a glide slope and localizer that they could track, then they could have shot this all the way down to the ground through the clouds and missed that fixed infrastructure. Some have speculated that it was a poorly designed, approach procedure, and the runway and that fixed infrastructure never should have allowed an automated approach, but, there's ways to do it. could have had a slightly higher glide slope, and we could have given this person something that would have saved their life.

Luka: 26:04

as tragic as it is, it's not a case for adopting a remotely controlled aircraft though, right? How do you draw a line from that to remote control of aircraft as being the answer, as opposed to other safety enhancing, tools and systems on the aircraft from, navigation, better situational awareness, et cetera.

Robert: 26:25

Well, the first system that we're developing is a fully automated aircraft that still has a pilot on board. our first step is not remote piloting. our first step is give the people in the cockpit, more sophisticated tools to prevent them from getting in those situations. And that does lead eventually to remote piloting, but there's a lot of other technologies that have to be added to the aircraft in order to enable remote piloting, such as air to air radar and more robust, communication systems.

Luka: 26:53

When you talk to your customers, part 135 mostly, I assume, do they admit that they have a safety problem? Is that what the business case revolves around?

Robert: 27:03

It's a substantial component, and for the reasons I mentioned earlier, there's, it comes first from a genuine desire to not put people at, a higher level of risk than necessary. If we can do something to prevent accidents and save lives, then damn it, that's what we need to do. and the customers that we talk to, the customers we spend a lot of time with, that's what they say. This is about saving lives. Having said that, there are other benefits that come along with it, once you move to remote piloting. I mentioned earlier, training, training becomes much simpler, much more straightforward. being able to reposition aircraft becomes much more straightforward. Being able to up gauge and down gauge the aircraft, flying aircraft that is appropriate for the mission or right sized for the mission is another benefit.

Luka: 27:47

tell us how your concept for remote operations, how that is fundamentally different to how the military is operating its MQ 1s, MQ 9s, or for that matter the QF aircraft series.

Robert: 28:02

our system is more like Global Hawk than the old Predator system. we don't have a joystick. there's no rudder pedals. there's no live stream video feed that lets you hand fly the plane down to the ground. our system Works, very much like an FMS, because it is an FMS that we've modified, and in fact, the first system that we're certifying, the system with the pilot in the plane, they're programming the aircraft through our flight management system. It's a slightly more, sophisticated flight plan than what you would do today. Today, you would go into a plane and you would program your enroute portion. maybe you would program your departure procedure and your approach procedure. But if you did those things, there would be potentially discontinuities in your flight path. And maybe your FMS will alert you to those discontinuities. A few systems actually do that. Many will just be happy to let you fly off, out over the ocean and never actually intersect your approach. And then while flying, air traffic control, and for other reasons, you may need to change altitudes, you may need to change headings, and that deviation from the flight plan is not managed by your FMS. So our FMS manages all of that for you, so you don't need to ever Disconnect the autopilot and hand fly the aircraft, we do it all. And we do that at a level of safety that far exceeds autopilots today. So our system on the ground now uses that exact same FMS. you're programming the aircraft, just like you would if you were in the plane using our system. What else is new? You also have a Traffic awareness, feature that's built in. So on top of our moving map, we plot traffic and our system provides certified or will one day be certified, today it's certifiable alerting and guidance, that maintains safe separation from other aircraft and will modify your flight plan for you. it also enables, voice communication to the aircraft. So there's a headset connected to our system and it works just like it does today. You dial in the frequency that you want to talk on. There's a push to talk button and your voice is streamed over satellite to the aircraft and then broadcast out the aircraft. So surrounding air traffic control, they hear you. Just the same way they would if you were sitting inside the plane. And conversely, you also hear everything just as if you were inside the plane. So we did it this way because, we're trying to do the simplest thing possible that can integrate into the existing airspace system. And I emphasize this because there's Folks out there that talk about, digitization of air traffic control, I think that's exciting. We're part of an RTCA group that's working on that, but I think that's a very long ways away. today The airspace works through charts, sectionals, AFDs, and people talking to people over amplitude modulated VHF. And so if you want to integrate UAS, that's the way you've got to do it.

Luka: 31:01

And Robert, given that there's so much experience across the General Atomics platforms, the Northrop Grumman's and some of the others. if there was a solid business case in the commercial market, Wouldn't you expect those same players to service part 135 operators?

Robert: 31:17

I don't know how much of this is public. so I'll speak in generalities, but major military drone manufacturer you've heard of, I know, has attempted to market, the system to commercial players and has attempted to push, their aircraft through an FAA certification process. And I, I think the short story there is, it's hard. it is really hard to take something that was not designed with certification in mind and modify it to be certifiable. the Certification process, and not just the certification process, but the set of standards that the FAA has recognized, that should be used for certification, take your systems engineering all the way back to requirements. and if you don't have the evidence going all the way back to the very, very early days, mapping the requirements for your system and why everything is exactly the way it is and precisely how you have tested everything to ensure that it meets those requirements, it's hard to go back and do that after the fact, and for those that are listening that are familiar with DO 178 software, there's common stories in there about how you can't just take something from a prototype and then snap your fingers and make it certifiable software. It just doesn't work that way. So it's hard for these military manufacturers to make that transition. millions of flight hours does not a certification case make. You need to be able to hit factors of safety in the 10 to the minus 8, 10 to the minus 9 realm. That means the acceptable failure rate is, depending upon the size of aircraft, one in a hundred million hours of flight, or one in a billion hours of flight. You can't fly a plane for a billion hours and say, well, we only had one accident, therefore, it's certifiable. I'm not smart enough, but there's a RAND paper that goes into this too, that says that even a billion hours isn't sufficient to show one in a billion. You got to go way beyond that in order to get any kind of statistical

Luka: 33:18

Right, right. I guess where I'm going, Robert, is when we look at the Enabling technology blocks that exist on these platforms. And there's been some recent examples where, one of these big OEMs, have integrated into civilian airspace in the UK, they do have a certifiable platform. So what's different in your system approach? On a fundamental technology block that stands in contrast to the approaches that you're seeing by these and other players,

Robert: 33:48

so you used a phrase there, fundamental technology block. And maybe I'll disappoint your listeners a little bit that I don't feel there's anything fundamentally different about what Reliable Robotics is doing from a technology perspective. What we're doing that's very different is our systems engineering approach. We are looking at this from day one with certification in mind, which means a thorough understanding of the regulatory construct for Highly automated aircraft and charting a path through the regulations, through existing published guidance by the FAA. Well, by the FAA, I was going to say and others, but we are nearly exclusively focused on the United States and building a system that complies with That set of regulations and guidance from day one. I think that's what makes us different. And when you approach it from that perspective, it leads you to make different technology decisions. for example, our air to air radar is not a military target tracking radar that consumes a tremendous amount of power, takes up a bunch of space and costs I don't know how much to manufacture. It's going to cost a lot less. it's going to meet exactly the requirements the FAA has, Published that they will accept for air to air radar for detect and avoid, through TSO C 211, 212, I believe, and nothing more. and that's going to result in a system that complies with those requirements right out of the gate and is probably an order of magnitude or more, less expensive than the systems that have been developed for military applications. I would also say the same on our communication system, our ground control station. Human factors is probably the biggest area in the FAA certification process where subjectivity creeps in, and that's where there are fewer standards to leverage, and what you really ultimately need to do is convince a number of subject matter experts in the FAA that the system does not Impose an undue burden on the pilot, and is consistent with human factors best practices, which change year to year, month to month, day to day. And that, that also needs to be designed from day one with a civilian aviation use case in mind. and that's where it's probably not a custom user interface that's point and click. it probably looks a lot more like the user interfaces that the FAA has accepted to date. and that's why our, design decisions track more towards a classic FMS and not something completely new.

Luka: 36:39

so whenever somebody makes a safety argument, the bar is really high, obviously. So how can, anyone pursuing autonomous flight Reliable Robotics included, make a convincing case that what you are developing is safe in every conceivable condition.

Robert: 36:59

Well, this just goes back to what we were just talking about. The way you show a system is safe is you follow the published guidance, that the FAA and EASA have long accepted. There are standards. There are system engineering process standards that have been around for many years that are used to develop the Boeings, the Airbuses, the Embraers, the Part 25 aircraft of the world. And the way that you convince the FAA and thus the public that it is safe as you follow those same processes. I fundamentally disagree with folks that say because it's autonomous, because it's new, we need to invent new processes. That's completely bogus. We have processes to do this. It's just, it's a very hard, it's an arduous path. That's how you do it. That's how you get to 10 to the minus 8, 10 to the minus 9 and beyond.

Luka: 37:51

So we talked about the fact that, the military systems have not been able to prove it after millions and millions of flight hours. What's your view on the path to getting there? In terms of, data collection, time, whatever metric that you are using to track your progress along that way.

Robert: 38:10

The metric that we use to track progress is how many regulations have we demonstrated compliance to so far? And there's processes for each. this is how certification works. You map out the regulations. you establish a certification basis, that says these are the regulations that are appropriate. You establish a means of compliance, an MOC. Fancy word, but just means how are you going to show that you meet the requirements. the, and then within each of that, there are processes. collect data, collecting data is a trigger phrase for me. You triggered me, it really irks me when people use that expression because I think, well, for what purpose? why are you collecting data? what system is this that you have where enough data on its functioning makes a safety argument? I'm not comfortable with a system where people say, well, because it worked for several thousand hours or because we did so many auto landings with it, it's safe. That's just bogus. That's not the way these processes work. That's how you get to 10 to the minus two. maybe.

Luka: 39:13

obviously data is important. Obviously showing a track record is important when you talk to the regulator. how do you then view this challenge of proving to the regulator that in the absence of data

Robert: 39:26

no, there is, I'm sorry. The collecting data is a trigger word for me because people talk about throwing their Autonomous plane up in the sky and building hours on it and collecting data so that they can further improve and modify the system. and then they, there's a double speak there where they often tie that to certification as if the FAA is going to say, Oh, you've flown 10, 000 hours, it must be safe. Okay, go for it. That's just irrelevant because it matters what it. You need to understand, like, well, what conditions did you operate the system in? did you just burn holes in the sky endlessly for 10, 000 hours? what were you really doing? I, we could take our plane and go burn holes in the sky over Mojave and perfectly clear weather with no other air traffic for 10, 000 hours, but what did we prove? Nothing. but

Luka: 40:12

it's configuration management. It is the right data logging. It's the right data management. obviously, it has to be structured in a way that resonates with the certification processes and all of the other applicable regulations. Absolutely. But, that's table stakes And so once that is understood, Talk about the development process, where you get comfortable and you get the regulators comfortable that you have achieved a certain, maturity level where you can field this to an aircraft and have confidence that whatever conditions they might encounter, the system will work and be safer than when the pilot is on board.

Robert: 40:47

I'll go back and explain the certification process. There are hundreds of regulations that impact what we're doing. and each regulation has, Congress passes laws, FAA creates regulations that are, in compliance with those laws. The FAA then publishes guidance. on how they expect applicants, like us, developers of aircraft systems, to comply with those regulations. So within that, they publish guidance on, I'm trying to think of an example. regulation. says that automated landing systems need to be certified. and then they publish, an advisory circular that says, here's how we, the FAA, believe you need to demonstrate that you meet the regulation for an auto land system. And that, that one line regulation or paragraph may expand into a 700 page document. That 700 page document may reference other advisory circulars. It may reference industry standards. Maybe RTCA has published some guidance on how to tune a radio system that uses radio navigation for the purposes of auto landing and what the performance requirements of that system should be. This all expands out into thousands and thousands of pages. We actually summed all of this up, recently. we're well over, I think it's 27, 000 pages of documents, that we've brought into our system that are all FAA standards and publications and guidance that the FAA has put forward and blessed as applicable for Complying with the regulations. And these are just the ones that impact our system. There's actually much, much more in aviation more broadly. And so you have to dig into each of these standards and pieces of guidance that the FAA has published to get answers to questions like, well, how many hours is enough? Or how many flights is enough? How many landings is enough? In the vast majority of these cases, our certification approval that we received earlier this year, maps to about, I would say, 90 95% ish to things the FAA has already put out there. The stuff that doesn't match is dealt with in issue papers, and there, we've either proposed our own standards. We've said, hey, if we were you, FAA, this is the standard. And we just made it up. This is what we think the standard should be for this type of system. And we've gone back and forth with the FAA. It's taken many years, but we've gotten them to agree that's a reasonable standard for this thing. And now that's the standard that we're going to follow. In other cases, we've taken existing standards and we've proposed tailoring them. We've said, for example, the auto landing standard for Part 25 aircraft has a lot of really good stuff in it, and we're just going to leverage that, most of it, 98 percent of it, for Part 23. There's a few areas that we don't think are applicable to Part 23. Don't ask me off the top of my head what they are. we got the FAA to agree to that. Okay, so now you get into, let's take Autoland for example, I don't know the numbers off the top of my head, but the Autoland set of standards, and this is hundreds of pages of documents that explain this, state that for you to get an automated landing system certified, you first have to build a very high fidelity simulation of your aircraft, and you have to show in simulation that the aircraft can hit essentially all possible, foreseeable, combinations of environmental conditions and aircraft configurations within whatever envelope you are intending to certify. So if we say that our system is good to 15 knots crosswind, then we need to show that for all density altitudes, and for all crosswinds, plus or minus 15, and all weight and balance configurations, and all aircraft configurations, and all possible, internal system failures that the aircraft could be experiencing, it will always bring the plane down to the ground within our performance window X feet laterally, Y feet, longitudinally, et cetera, and we're never going to exceed the G load limits of the aircraft. then you take the simulation, then you, from that, you say, okay, well, how do we verify that the simulation matches reality? And you have to choose test points in the real world that give you confidence that the sim is representative of the real aircraft. So you may want to pick the worst case aft CG configuration at the worst case density altitude at the worst case crosswind component, and then you have to go find a place in the United States to go do this. And that's why you go out to a place like Casper, Wyoming or whatever, and you stay out there for a couple of weeks until you hit that test point. And then you hit all your test points and say, okay, boom. Now, the simulation is good. We comply with all the Autoland requirements.

Luka: 45:40

in developing autonomous systems, you can broadly bucket into data, algorithm development, and testing, which one of those three do you see as the most challenging?

Robert: 45:50

Say that again, algorithm development and what?

Luka: 45:52

broadly data, and in that it might be data collection, data annotation, warehousing management, so broadly the data bucket, then it's the algorithm development, and then it's the testing validation and verification. So if we can high level those three categories, where is your biggest focus now? what's the long pole in the tent? and where does a startup have an advantage over an incumbent?

Robert: 46:17

can I see what's behind door number four? I, I think, I wouldn't pick either of these. I would go to the certification basis, I think is the biggest challenge. And that's the area where a startup like us can have the biggest positive impact. And that's why we started the company. I don't think, I don't think collecting data is, a huge technical challenge. that's solvable. Algorithms, while complicated, are just algorithms, and that's just work. and the testing is just work. It just, it's something that you just have to do. the really hard part, the most challenging task in all of this is to figure out how the autonomy maps specifically to the regulations,

Luka: 47:00

while we're on the topic of Algos, can you talk about the software that's behind your autonomous features? Is it rule based programming? Is it condition based? Is it machine learning? Something else?

Robert: 47:12

This is very straightforward, normal software.

Luka: 47:16

So rule based?

Robert: 47:18

it's, this is, these, it's procedural logic, it's if statements, it's conditional logic, um, it's not machine learning.

Luka: 47:25

so in that case, you're a pilot. Most of us are pilots on the call. how much of the situations that one may encounter in flight do you feel that your software can be programmed to appropriately address?

Robert: 47:38

If we can anticipate it, and this gets back to the processes that I was talking about earlier, there's a safety assurance process that's used in part 25 aircraft that we're using for the development of our system, where you enumerate all of the possible things you can imagine that will go wrong and you decide how will the system aircraft and pilot manage it and how to prevent a potentially hazardous or catastrophic outcome as a result. We're going through those exact same processes here. The difference is, we just don't have the pilot inside the aircraft to act as part of the system, if that makes sense. And it's hard. it's a lot more work when you don't have the human in the cockpit as the final node in your fault tree, to save the day and you don't get to defer the problem to training, then that means you need to implement an algorithm to, to solve it. And if we can imagine it, then we need to implement code or some feature of the system to mitigate that hazard. If you can imagine a hazard, we need to mitigate it. Period. That's how it works. Or we need to convince ourselves that it's like extremely improbable, like beyond 10 to the minus nine.

Luka: 48:56

a lot of these, situations you cannot imagine ahead of time. So when is good enough, both good and enough, who makes that call? goes back to to how do you with authority stand in front of the operators, the people on the ground, the regulators, and say, we're good.

Robert: 49:12

yeah, maybe I'm not getting through. It's following the processes that the regulator has approved. That's how you do it.

Luka: 49:20

So let's talk about cert, the processes, and if you don't mind explaining your certification path, and I'll just reference, one of your competitors, I think in 2021, petitioned the FAA for waivers for, a remote operation and was denied. how are you planning to avoid this situation? Can you give us a little bit more detail about your cert plans?

Robert: 49:41

Well, we don't intend to get in that situation. I am not surprised that their request was denied. there's a reason we have the certification process in the United States. And, I don't think they made a compelling enough argument to, short circuit that certification process. and so what we're intending to do is to follow the certification process. Our system is an STC that goes on top of the existing aircraft that's already type certificated, modifies it to enable remote piloting, and then the Part 135 air carrier certificate and the operation specifications associated with it need to be modified to support the STC. Much like today, if you have a 135 and you want to I don't know, fly CAT 2 ILS approaches, you have to write that into your OPS spec. You need to demonstrate to the FAA that you have sufficient training procedures to, teach your pilots how to run a CAT 2. they'll be the same kind of thing with our system where if you want to run our STC in your operation, you're going to have to write that into the OPS spec and you're going to have to coordinate with the FAA and get permission to do that. So for those of you unfamiliar with how airlines work, certifying an airline is just as complicated, if not more complicated, than certification of the on board equipment, the aircraft itself. the FAA gets very deep into how an airline runs itself, and the procedures that airlines use. There's a few obvious things like, pilot training that they get very deep into, because all airlines have chief pilots and chief pilots are responsible for, ensuring that everyone is sufficiently trained and there's standards that they need to follow. Safety management systems is another aspect that the FAA gets very deep into, and how, airline operators build and maintain safety culture is a big part of that. Maintenance is another obvious feature that the FAA gets very deep into.

Jim: 51:42

All right. Hey, Robert. so let's take a breath and let's go in a different direction here. Great discussion. How big is your addressable market today?

Robert: 51:50

Well, are we talking about today right now? Because I don't have a product yet. I don't have a certified

Jim: 51:54

you had a product today. How big is the addressable market?

Robert: 51:57

Well, our system is designed to be aircraft agnostic. And our intent is to take this STC, that we're first working on in the Caravan, and then modify it so that it can be put onto other aircraft. It'll be somewhat straightforward to do this for other Part 23 aircraft, a bit more complicated to do it for Part 25. But if we're including, Part 25, in your question, then the addressable market is huge. It's basically all of aviation. aviation is close to a, what, a trillion dollar industry now, drives four and a half percent ish of global GDP. and we think that we'll be able to access a very large portion of that. I should say to our business model, because I think this is important. is not just to provide the equipment. We don't want this to be a one time sale, but we're also providing all of the training and infrastructure necessary to install, deploy, maintain, and operate the autonomy as well.

Jim: 53:02

What are people willing to pay for today? And why? By market. And how much of it is safety driven? And if it's not safety driven, what's the key value proposition?

Robert: 53:15

It's safety driven. We were talking about that earlier.

Jim: 53:17

Even 135 and 121, if there's a market with a 121 today, you're saying that the safety driven value is the number one driving need for autonomy.

Robert: 53:30

That's my position and I'm sticking to it.

Jim: 53:32

Okay. And five years from now, you define the size of the market by the trillions of the size of the aviation market. But realistically, Let's say you're in front of a group of investors and they're thinking, is this a company, is this a space

Robert: 53:48

Oh, you think I wasn't being realistic? All right, okay.

Jim: 53:51

And if this is a space I want to invest in and this is a company I want to invest in, what are people willing to pay for in five years by market segment for autonomy?

Robert: 54:01

The most blunt way to put this that I think simplifies it, because yeah a trillion dollars, is a huge number. Let's see for your audience, let me decide how to answer this, because well, first, the most naive way to look at it is, okay, you're automating the aircraft. That's a function currently performed by the pilots. Well, what portion do, pilots contribute to the cost of aviation? And the answer to that is on part 25 aircrafts, it's close to 20 percent and on part 135, it's close to 30%. In some cases, it's 35 percent or more. A naive way to look at it is, okay, if all we do is reduce piloting costs, then it's 20 percent of the 800 something billion dollar, Part 121 world, which is a big number.

Luka: 54:48

But Robert, it's not that autonomy would replace the pilot, it would just move it, right? There would be efficiencies in centralizing these pilot services so you can spread them around the network and make the network more efficient, that is for sure, but it wouldn't be a one for one replacement or removal of the pilot.

Peter: 55:04

I would think that autonomy is necessary for replacing the pilot in Part 121, but it's not sufficient. There are a lot of other things roles that the pilot is playing in the airline and in the operation of that flight, right?

Robert: 55:17

Yeah, and I don't want to say that it diminishes that cost to zero, but it is a substantial reduction. some of that I think still needs to be tested, and we're working towards testing it as soon as we have our STC. We'll be able to, to go test out this and validate the unit economics, but it is a substantial, we are anticipating a substantial savings Right out of the gate. just by having the pilot sit in a control center, the effective utilization of that pilot goes way up. this is especially true in 135 operations, where, pilots typically fly less than two hours a day. And having them in a control center, they could fly a lot more. But then, that, I don't think that's looking deep enough that's, as I said, the naive way to look at it, the very, the oversimplification, where this gets quite a bit more nuanced, is when you look at how you can change the way your operations work, now that the pilot is in the control center. And the first thing there that we hear from operators that's most important to them is repositioning. Freeing the aircraft from the pilot, and this could even just be a completely empty plane that just needs to deadhead somewhere, a repositioning flight, would drive huge operational efficiencies. determining the value of this is complex. It depends on an operator to operator basis. But the feedback that we've received from many of them is that this is a substantial driver for them. The second is being able to, right size the aircraft, up gauge or down gauge. if you're remote piloting and you're flying an airplane that has a much more simplified interface, and the automation is taking care of all of the aircraft's specific details for you then, you don't necessarily need to be specifically trained, aircraft to aircraft. And so you can switch flying, one morning you might fly a Cessna Caravan, and then the next you might fly something like a Cessna SkyCourier. You can't do that today. It's not considered safe. It's not safe to have people trained up to fly multiple aircraft simultaneously and be jumping around back and forth because it's hard to maintain proficiency, across different aircraft types. maybe some pilots would disagree with me. I'm going to take that position, and there's many operators that take that position, even within, like, variations of the Airbus fleet. there's some operators I know, they will not let their 320 pilots that are trained to fly one avionics configuration fly the other because they don't consider it safe. So, remote piloting, you can now do that. and that, it also enables a huge amount of operational efficiency, being able to right size the aircraft.

Peter: 58:01

And what do you see as the time horizon for Reliable to be able to realize that vision? I want to tie back to Jim's question, on that five year time horizon, unlocking value for the company. You've been focused on the Caravan of which, roughly 3, 000 have been manufactured in its lifetime. a subset of those are in a type of operation that falls within your target market. But what does it take to expand the footprint of your product via STC into other platforms? How much time does that take? And does all of this happen within the time horizon that Jim laid out in his question, as we, we drill down from the top level market size down to the direct market that you can capture in five years.

Robert: 58:46

The key inflection point for us is, once we've got this system certified and operating on one aircraft type because of our focus on certification and the certification basis and all of the tools that we've developed to enable these, processes and analyses to be performed, don't anticipate it to take all that much time to transition the technology to subsequent aircraft. So I see an inflection point for us where once this is certified on one aircraft, boom, we can quickly scale to modify it and deploy it to other aircraft. So the time horizon question in my view gated or the critical path is on getting this very first system certified. That's the focus. so then you're going to ask me, well, when do you think that's going to happen? And it's hard to say. for all of the reasons we've been talking about previously, these systems engineering processes are very complicated and when you start digging into the details, you find things and it's hard to predict when you're going to be done. I can tell you when I'd like to be done. And I can tell you the dates that we have in mind, the team is tracking. but it's challenging to predict because nobody has ever done this before.

Peter: 1:00:03

Is your development path of a radar capability, is that on the critical path to what you're describing here?

Robert: 1:00:11

The radar itself, the hardware, the software, the signal processing, no, the certification basis, yes, that is a component. I wouldn't say that's the, the biggest thing that's keeping me up at night, but it's certainly, it's a, an important component. The FAA has accepted, a set of standards for air to air radar. To my knowledge, nobody has yet driven that all the way through the certification process. We will likely be the first, and we're going to encounter things when we do that, and it's hard to predict how long it's going to take.

Peter: 1:00:42

And so when you, embarked on this project and you did the breakdown of, okay, well, are there ways that we can acquire this data? And I know that you have, done a lot of thinking on that. are there ways that we can lean on the supply chain to get a radar solution that meets our needs versus, what is the, time and investment and risk involved in developing our own? walk the audience through what those choices looked like and how you settled on this path?

Robert: 1:01:13

Yeah, this is a recurring theme or nightmare of my career. we, we dealt with a lot of this at, previous companies that I'd worked at. you're trying to do something new. You're trying to put a product out in the world that nobody has ever done before. And you want to do it as quickly as you can. And so your mind goes immediately to, okay, who can we partner with? who can we get this component from? Radar was one where we had been trying for a very long time, to convince another set of developers, I won't mention any names, but companies you've heard of, to go build, a radar that meets our requirements, or rather meets the requirements the FAA has already published. And we've said, if you build this thing, we will buy it. And we've got customers, and this is how many units, you can expect us to purchase from you on this rough time horizon. And this has happened with us, with radars. It's happened, with actuators. It's happened with flight computers, with IO concentrator computers. It's happened with, our communication system. There's a couple of these key components that we didn't initially set out to build this ourselves, but we ended up deciding to bring it in house because we either, couldn't get anybody else to build it for us, period, we couldn't get anybody else to build it on a reasonable timeline, or we couldn't get anybody else to build it at a reasonable cost. And with the radar in particular, that was one where we went back and forth for a very long time, and eventually it just became clear that we're burning more time trying to convince this other group to go after this billion dollar market opportunity when we could just be spending this time building it ourselves. And at least then we build ourselves, when we encounter issues deploying it, which we inevitably will, it will at least be a misery of our own making.

Peter: 1:03:14

So let's dig in on that project. what are the deficiencies that you are trying to address with this in house R& D to build a radar that meets your needs? What are the gaps between what's available and what you want? And what do you see as the resources required to pull it off? Radar is not like a weekend project, right? It is an industry all in of itself.

Robert: 1:03:36

Oh, it's not? Oh,

Peter: 1:03:37

there's actually a

Robert: 1:03:38

I was told it was going to be done next weekend.

Peter: 1:03:40

there's a lot of rapid innovation that's going on in that space right now. And so how do you look at that, risk versus reward profile?

Robert: 1:03:49

Well, first one realization was that radar, the components to build a more sophisticated radar have become more widely available in recent years. If we were having this conversation five years ago, I think you would be on much firmer ground to say that we're crazy. But now much of the signal processing technology, has become commoditized through deployment of global 5G networks. Antenna manufacturing is still, really tricky. and there are a few areas where we are still partnering, but, we're now the main integrator of the component. I'm not answering your question. You went to, What was the first part of your question again?

Peter: 1:04:30

The first part of the question was, the required resources and the gaps that you're trying to close between what's available in the market and what you need to have a radar solution that works in the National Airspace System for autonomous flight. And then what are the resources required to pull off that R& D project and what is the, risk profile of that undertaking?

Robert: 1:04:55

Roughly speaking, the air to air radar requirements for an autonomous aircraft, the reason you're putting a radar in there is to comply with 91.113 see and avoid requirements. And RTCA, a major standards body has been publishing, Minimum Operating Performance Standards for air to air radar systems for several years, the FAA has accepted, some of this work through, the TSO publications that they've put out. the requirements for those radar systems are largely a function of encounter speed, and within that, you have a range and resolution and tracking requirements, so you need to be able to see something at a certain range, you need to be able to infer its velocity at a certain amount of precision, and then you need to be able to track it, at some particular rate, and you need to be able to track a certain number of things, and you also need, a certain amount of visibility, vertically and horizontally. so those are the requirements. so this might be, I don't know, you might need to see out to 6 or 7 nautical miles, for example, and you may need to be able to see another aircraft that's moving 400 knots relative to you, for example. that then, it just goes to radar physics, and from radar physics, you conclude, okay, well, if we're going to use this radio frequency spectrum, then we need something this size and this power, and we need this many antenna elements, and then you go out to the market, now to answer your question, you look at, okay, who has a radar that meets these requirements, or is close, or could modify something to meet the requirements. In there, radar developers, fall into three buckets. you've got existing aviation certified radars that are designed for weather applications. those have incredible range, but the resolution and tracking is not sufficient. You've got military radars, which definitely have the range, definitely have the resolution and tracking capability, but they are tremendously expensive, they're huge, and they consume a lot of power. And it's hard to imagine how installation of some of those systems would be practical at scale on a Part 23 aircraft. And then you've got, sort of new entrant radar manufacturers that are using clever algorithms to do things at lower power with less, radar reflectivity information, and it's tricky to see how they modify those systems and scale them up, because they don't currently meet all of the requirements. And so that, that investigation led us down the path of, okay, let's try and convince one of these folks that are operating in the radar development space today to modify their system to meet the FAA's requirements. And that's ultimately what led us to where we are today.

Luka: 1:07:49

What technologies do you foresee being an important and necessary onboard part of an aircraft, as opposed to, being offered as a service? What do you think is the right combination for the mission that you are pursuing?

Robert: 1:08:04

Well, I think what you're question, getting at here, is what are the key technology components that are necessary to enable autonomy?

Luka: 1:08:10

Especially in the context of detect and avoid, obviously you can have a radar onboard, which you are pursuing, but you can also, and you've had some experience, so maybe you can share some of the, insight with the audience here in terms of, asking for primary surveillance data from the FAA radar feeds. how do you think the market evolves around what ultimately gets equipped onboard versus what gets offered as a service. And then how does that change depending on the mission?

Robert: 1:08:40

Okay, I see what you're getting at. I think of the technology that enables autonomy in two different categories. Step one of autonomy is to have an autonomous aircraft. an aircraft that can manage itself through all phases of flight. The second thing you need is you need an airspace integration solution that integrates that autonomous aircraft into the system that we have today. And I see these as distinctly different things. Autonomy, or the airborne automation component, consists of all of the things that you would need to do automated taxi, takeoff landing. You need also a high reliability vehicle management system that can manage the fuel, circuit breakers, and all of the other subsystems that exist on aircraft, to keep the aircraft in a safe configuration. And now once you have that, once you have an aircraft that's capable of flying itself through all phases of flight, then it becomes a question of, okay, how do you integrate this into the airspace? And that's where you need to comply with 91.113 see and avoid requirements. And that's how you end up with the air to air radar. You need to communicate with air traffic control and surrounding aircraft. That's how you end up with the communication system. We have a dual redundant satellite communication system that we're working on certifying. And then finally, you need the ground control station, and to satisfy all of the human factors requirements associated with, with that interface.

Luka: 1:10:11

but specifically see and avoid, where do you see the, the market converging, onboard radars or, surveillance data that is provided as a service.

Robert: 1:10:21

I think it's actually long term it's going to be a fusion of many things, because air to air radar is good at certain things. Primary surveillance radar is good at other things. camera systems are good for other types of things. and so I think we'll see, a fusion. Our path that we're taking, now, and what we believe will first be accepted by the FAA for aircraft the size of the Caravan will be an air to air radar system, and that will meet the requirements and allow us to do operations.

Luka: 1:10:53

Can you tell us about your experience requesting primary surveillance data from the FAA?

Robert: 1:10:58

yeah, so primary surveillance radar, for those not familiar, these are radar systems that were initially put together by, a lot of them were built by the U S air force, for national security purposes. And then the data was released to the FAA for air traffic control purposes. They've since been augmented with transponder data feeds, and the combination of the primary radar with the transponder feeds, is frequently referred to as secondary surveillance radar. and the pictures of air traffic controllers, the screens that they're looking at is, is a terminal that shows you a fusion of primary and secondary, surveillance radar. Or frequently just secondary, but, they have the capability to see primary as well. Primary surveillance radar, in the United States, and there's some publications that the Air Force has put out there on this, is pretty incredible. There are very few places in U. S. airspace that are not covered by primary surveillance radar, and for the types of operations that we need to do with small cargo aircraft like the Caravan, it's possible to fly the Caravan and stay within primary surveillance airspace with sufficient margin to see all of the things around that aircraft. And we've shown through a variety of simulations. There was some work that we did with NASA several years ago on this. we've collaborated with some folks at MIT Lincoln Labs, on this as well. We've shown that you can actually use primary surveillance radar to solve see and avoid in certain airspaces in the U. S., not all, but in many of the ones that are economically relevant for our use case.

Peter: 1:12:40

And Robert, does that account for the, the update frequency of primary surveillance radar and the latency in getting that data to the aircraft?

Robert: 1:12:47

it's really a function, too, of, well, how it gets back to what I was talking earlier, about air to air radar requirements. it matters what air speeds do you expect to see encounters and how much separation buffer do you want the system to have. And so in areas where the update rate and resolution is lower, it just means you need to fly further away from other things that the radar picks up in order to maintain sufficient margins.

Peter: 1:13:14

What is the update rate?

Robert: 1:13:16

That I don't know off the top of my head.

Jim: 1:13:18

Robert, what's the likelihood you would get access to primary radar in the next couple of years at scale?

Robert: 1:13:24

Well, so we actually had access many years ago. I think we talked about this in that article in The Air Current. and that was partly how we demonstrated the efficacy of PSR for solving this problem. and then we went to go do another set of test demonstrations and our access was revoked and the message that we got back was that too many other groups had taken notice of what we did. Word got out and so a lot of other people started requesting access and then somebody raised the specter of national security.

Jim: 1:13:57

without a doubt. it took 20 years for secondary radar information to be provided to the public. I can't imagine what it's going to take for primary radar data to be released. Again, at scale. testing is one thing, but at scale is another.

Robert: 1:14:11

But we anticipated this and I, let me, I don't want to let that drop. We care about national security. I care about national security too. We anticipated this and we proposed a number of processes that the FAA and the DoD could use to secure the data. We even went so far as to, suggests that the data only be accessed within locked facilities that are access controlled and follow a variety of NIST standards for cybersecurity and none of that was accepted. And my understanding is, the FAA is continuing to look at this. They are making progress with the DOD on how to release the data. I've even had executives in the FAA tell me that they think it is in the interest of national security to release the data because if they don't, then it means private companies, potentially foreign owned private companies, might stand up their own radar networks and have better surveillance in the airspace than the United States government does.

Jim: 1:15:11

Stand up their own primary radar within the United States, foreign governments?

Robert: 1:15:15

Foreign owned companies may do this, and I would say that there are private companies today that already have stood up their own radar systems at airports. This is

Jim: 1:15:26

millimeter wave and the like, but not at scale, not across the country. Hey, let me, I,

Robert: 1:15:30

talking about doing it at scale.

Jim: 1:15:31

How critical is the primary radar to your airspace integration project? And how critical is the airspace integration project to you fulfilling your financial requirements over the next couple of years?

Robert: 1:15:45

Primary is now secondary, if I can make a joke. We've shifted our focus towards air to air radar. We are still, lobbying Congress and educating members of Congress and the FAA on the safety benefits of releasing primary surveillance radar. I see it as a good way to augment air to air radar. It's strictly better to have it out there. I'll just leave it at that.

Luka: 1:16:10

Robert, if I'm not mistaken, today you run a small fleet of Caravans. How much do you want to grow this aspect of the business? Is it a growth area for you or a necessary short term step?

Robert: 1:16:21

Necessary short term step. We started this airline. We view it as an incubator and this gets to what we were talking about earlier with the certification of the airborne component and modifying the Ops spec for the airline. We decided to bring that portion of it in house so that we can modify the OpsSpec ourselves first in a very small airline, relatively small airline, I should say. And once we figure out how to do this and get the FAA to approve it, our intention is to then license it out so that other airlines can benefit from it.

Luka: 1:16:57

And what do you see as the optimal position to occupy along the value chain? What are the pros and cons for each? Whether you're an OEM, an operator, an avionics supplier, an airframer, in this world of autonomy?

Robert: 1:17:12

I think there's places for everybody to, to get wins. OEMs are going to benefit because autonomous versions of the aircraft are going to sell much more than other versions. Maintenance providers, FBOs, are going to benefit because there'll be different maintenance challenges. We were talking earlier about ground handling complexities with remote piloting and how we're going to figure all that out. And fixed base operators and others are going to benefit from that. Component suppliers are going to benefit by pushing the technology out. Airlines, obviously, are going to benefit through operational efficiencies. Where we see ourselves fitting in is we want to provide a few of the core components, not all, but a few of the core pieces of equipment, and we want to provide the certification services to get the equipment put into use. And I'm talking about the airborne component as well as the Airline operations component. And then there will also need to be ongoing, services, communication services, as well as detect and avoid and other cloud based, aircraft management components that you'll need in order to be able to run this technology at scale. And I think that's an area where, we'll be able to provide value for our customers too.

Peter: 1:18:26

Robert, do you see companies that are building autonomous flight technology in the small UAS space entering this value chain and being a player in the coming years? On the one hand, they are developing very rapidly. They're able to do tighter iterations. They're able to fly a lot more hours. On the other hand, they're, building their innovations in a different airspace regime, below 400 feet, different requirements for that. But I think one could argue that some of the technologies that are being matured there are advancing very rapidly and, have relevance in the rest of the airspace system. So when you look out ahead, where do you see those manifesting in the value chain?

Robert: 1:19:14

Well, I'd go back to what we were talking earlier about military UAS and applicability. I'd give you almost the same answer. in order to meet the certification requirements, the regulatory requirements, you've got to take it all the way back, to the beginning. And while I'm sure there are lessons being learned in the small UAS space and, we're benefiting from work that's been done there, early work with Beyond Visual Line of Sight, the core components and the regulatory basis, I don't see that carrying forward. It'd be like saying an autonomous fork truck in a warehouse is going to turn into a self driving car. Sure, like, there's aspects of it that we'll carry over, but it's just a very different design domain.

Peter: 1:20:00

So you don't see potential of the safety case and the data built up in that domain being extended into other types of flight or other parts of the airspace system, that's really just not going to be a factor.

Robert: 1:20:13

I hate to give you a firm no, because there may be an element to this that I'm missing, but, I, I struggle drawing a straight line between the two. the Part 107, uh, 44807 exemption space, It's just very different from part 23 aircraft and part 25 aircraft. It's night and day.

Peter: 1:20:30

Does part 108 from, what we are hearing, of its development, change that equation in any way, or have you heard anything on that front?

Robert: 1:20:39

I haven't heard anything. no, I don't think that's something I don't, I couldn't comment intelligently on that.

Jim: 1:20:44

Robert, what question were you dying for us to ask that we haven't asked yet?

Robert: 1:20:48

Oh, we covered a lot of stuff. I think I pivoted a few of your questions into the talking points that I wanted to hit.

Jim: 1:20:54

You're like the Pete Rose of podcasts.

Robert: 1:20:56

This was a lot of fun. Thank you for tolerating me.

Jim: 1:20:58

The other thing I would have loved to have gotten into, but I'm looking at time is your background with, Tesla and SpaceX and, just give us a really brief answer. with you as an executive, with you as a technologist, what do you do different than others because you've experienced with Tesla and SpaceX?

Robert: 1:21:17

Well, I think the biggest thing that I've taken away from those experiences is a hesitation to, accept the status quo. Maybe that sounds a little too cliche. There's a Werner von Braun quote that I love, and I think he said this sometime around the launch of the Saturn V. Wernher von Braun was the designer of the Saturn V. and he said, I've learned to use the word impossible with the greatest of caution. And there have been a few points in my career working at those companies that you mentioned, SpaceX, Tesla, where it hit me for a moment, I felt like I understand what he means when he said that.

Jim: 1:21:53

Right.

Robert: 1:21:54

This company, Reliable, we're doing things that many people told us before we started were impossible. We're still told it's impossible, and I just refuse to accept that. We will find a way. I think we need to do this. I believe this is the next big step that humanity needs to take. I think we need autonomous planes to further connect people and connect the planet. I think this is going to be a bigger transition for humanity than the jet age was. the jet age shrunk the planet, and autonomy is going to shrink it even more.

Jim: 1:22:25

and, don't pivot off this question.

Robert: 1:22:27

Okay, alright, I'll try.

Jim: 1:22:29

is your biggest concern in the next couple of years, your ability to execute on the technology or the market's willingness to pay for that technology?

Robert: 1:22:39

I definitely don't think it's the market's willingness to pay. I don't think that's going to be a problem. However, I would pivot slightly on the technology that I really think that certification is the challenge. Certification is the challenge. We need the FAA to do this. This takes a massive public private partnership, and, I know I'm the one you're speaking with, but, and I'm representing Reliable Robotics, but I also feel like, the FAA has a big role to play here. And it's of crucial importance that Americans especially take note of this and inform members of Congress that the FAA needs support to do this. If we want the United States to continue to be a leader in aviation, we need to be putting more resources into the FAA. We need that FAA reauthorization bill done.

Jim: 1:23:24

And with all of your experience in running small companies and with working with large, great companies, what advice would you give to our entrepreneurs on the line?

Robert: 1:23:33

First, build a product that people want to buy. I guess this is just generic advice for anybody who's an entrepreneur, but, make sure you stay customer focused. I see a lot of companies that are working in the aviation space and I'm not convinced that they're building something that customers actually want. So stay customer focused. Second, I would say hire the absolute best team you can. Do not compromise on hiring.

Jim: 1:23:59

As you did with Brandon. And

Robert: 1:24:02

Thank you. Raise a lot of money, raise as much money as you can, you won't regret it. and that's probably it.

Jim: 1:24:08

how would you like to wrap up the podcast?

Robert: 1:24:10

I feel like you guys have been pretty generous with me and you've let me pivot off of a number of things to hit all the talking points I wanted to hit. I can't think of anything else.

Jim: 1:24:18

That's a great summary. Because you're thinking what we're thinking. great. Anything else, guys,

Luka: 1:24:23

I feel like there should be a part two coming up

Robert: 1:24:26

Oh dear. Yeah, when he, when Robert actually answers the questions.

Peter: 1:24:29

We shall try again.

Robert: 1:24:30

yeah.

Jim: 1:24:31

Right, right, right.

Luka: 1:24:32

We'll

Jim: 1:24:32

Thanks

Robert: 1:24:33

was a lot of fun and I, seriously, I had a great time. I really appreciate the opportunity. I appreciate the thoughtful questions, and yeah, not all of these were easy, but thank you.

Luka: 1:24:43

Thank you, Robert.

Peter: 1:24:44

Thanks, Robert.