Jennifer Taylor 

Ladies and gentlemen please welcome our what's next for vehicle automation panel

 
Finch Fulton 

Okay, well Hello, everybody. It's wonderful to be here with you all today. My name is Finch Fulton. I am the Deputy Assistant Secretary for transportation policy at the United States Department of Transportation. And I'm proud to be here before you on behalf of Secretary Elaine Chao. We have a terrific panel in front of us today. We've got lots of interesting things to talk about. But using the moderators privilege, I will highlight and underscore the Secretary's 1130 announcement of automated vehicles four point O, ensuring American leadership and automated vehicle technologies. This document is the next step and ensuring we have a whole of government approach to getting the benefits that we think automated vehicles can bring to the public name, namely safety, the efficiency of the system and the economic benefits that can come of it, but also very importantly, the accessibility and the opportunity These technologies can bring for the elderly or people with all types of disabilities. So these are things that we're working on actively and things we're very interested in excited about. But it all comes down to safety and how we think about safety and how we incorporate safety into everything we do. So I would let my panelists give an introduction of themselves assuming even though this is a very savvy audience, that not everyone knows exactly who they are, what they're doing and some of the interesting and innovative work they're doing. Please let us know who you are, what you're doing and why it is that your company is so innovative. Sure, no, sir.

 
Sterling Anderson 

Yes, please. Sure. Sterling Anderson, co founder and chief product Officer of Aurora. We are a self driving development company based in Palo Alto, California, in Pittsburgh, Pennsylvania. We've been around for about three years. And we're working with a number of different partners to bring the benefits of self driving technology to market safely, quickly and broadly.

 
Josh Switkes 

Excellent. Good afternoon, everyone. Josh Swick is founder and president of peloton technology. At CES. I feel like I have to make it clear. We're not the exercise bike company. We're a connected truck. Automation company. So specifically we leverage sensors combined with vehicle to vehicle communication, to bring automation to the the freight and trucking worlds bringing improvements, the safety in fuel efficiency and in labor productivity. We're based in Mountain View California. We've been around since about 2013. So a little bit longer than than some companies although Sterling was being modest. He's been in the industry much longer than that. And excited to be here with you all today.

 
Debbie Hersman 

First Finch, congratulations on the AV 4.0. Happy New Year everybody. My name is Debbie Hersman, I work for waymo and I'm excited to celebrate my one year anniversary as the chief safety officer at way mo Our mission is to make it easy to move people and things around and our CEO john craftchick announced on Monday that we've just achieved 20 million self driving miles this past year, so excited to talk about The other panelists today.

 
Ralph Lauxmann 

So hi, I'm Ralph Lauxmann working for Continental, a big automotive supplier, traditional automotive supplier company, also engaged in road safety all the time. And that's the reason why we are also engaged in automation of the traffic, automation of the cars. And we're developing that worldwide. So we have development departments in Europe, in the US and Japan and in China and not newly also in Singapore. So that we develop the mobility scenarios of the future based on the experience that we gain everywhere in the world. And one of the main motivations to do that is that we increase the safety on the road further and, and build into a level so that maybe no fatalities happens anymore.

 
Finch Fulton 

We're all working towards the goal of zero fatalities. But it's interesting because I heard you all highlight safety in your conversations about what your company Focus on and how you're engaging with these new technologies. For us, obviously, safety is our top priority, you see that carry through for the whole of federal government approach and how the United States government will be engaging with these technologies. But what we want to do is as we look to achieve greater levels of safety, not be top down, and we want to be tech neutral, allowing the best safety approaches and the best technologies to come forward in the way that makes the most sense for each company. So as you approach safety, we have different ways to approach safety, using connectivity using some of the systems that you've got your different approaches, can you highlight your company's approach to the safety of these evolving technologies? We want to go first? Sure,

 
Sterling Anderson 

yeah. Safety is very multi dimensional in this space, right. We think about it at the outset. We have to create kind of a foundation and a culture of safety. Right and it Aurora. A big part of that was when we founded the company, we made it clear with our employees. It anyone at the company from founders down to whomever should they feel uncomfortable is, should ground the fleet globally across the company. This is something that happens with some frequency. And it requires two of the three founders in a convening kind of have the appropriate people to lift that and grounding. That's part of the culture we've tried to imbue kind of the company with so that everybody understands this is kind of first and foremost thing that we did is we've been very particular about how in when we lift various policies engagements on the road, we have to recognize that members of the public did not opt in to be guinea pigs in tests. And as a result until and unless we are confident that our system is sufficiently safe that we can safely test around them, we will not force those people into a situation where they are taking a chance with whether our vehicle does the right thing. And so, what we've developed on the back end is a strong virtual test suite that allows us with manual driving data, which is say data collected with far lower risk than autonomous driving data to evaluate and in fact develop much more quickly than we otherwise might have. And one of the things that we're excited about is in this push from online toward more offline virtual test environments, we've just achieved what we believe is a bit of a milestone for us as companies in 2019. This next generation development has allowed us to to move much, much more quickly developmentally than we have in previous years, while driving fewer miles on road than we did in 2018. And we think that for us, that's kind of the holy grail of development, we can move past the traditional approach to just bulk miles collected on road autonomously toward a more offline more robust kind of thought process.

 
Finch Fulton 

So how do you go through the process? And this is something I'm going to look to Debbie to weigh in to as well. How do you go through that process of expanding the types of operations that you feel comfortable in that all of your employees are together feeling comfortable putting out onto the road? What does that process look like for either way, Mo, or or? Sure.

 
Debbie Hersman 

Sure. So I think I think you know, it's Some of the issues that Sterling raised are about just different approaches that companies are taking in, I'd say at way mo for 11 years now they've been working on this technology. And it really is a step by step approach to rolling out the technology, doing it very slowly, very methodically, making sure that we have rigorous requirements, reviews, data driven decisions that we're making about where we drive, how we drive, how much we drive. And so I'd say that for us, we actually see real strength and having multiple ways to evaluate our performance. Definitely, using simulation is a very powerful tool. And so we're driven over 10 billion miles and simulation, but we also see structured testing and controlled track testing is a really important component to to look at edge cases to make sure that we understand things that how the car performs in every scenario. And I'd say Lastly, though, on the road to Testing is a huge part of why I think that we feel comfortable putting our vehicles out on the road, you have to operate in the environment before you go fully autonomous in the environment. We're operating in the environment. regularly, we're collecting those miles. And when we decide to pull the driver, it's with confidence that we know what we're going to see. And so I would say we're learning things through manual through those miles that we're driving, and through those automated miles that we're driving, that we did not learn in simulation. And so I think it's really important to understand that there have many different approaches. And I think we really want to recognize that everyone comes to this with strengths and they have the ability to kind of put forward what they think the best approaches But for us, I would say we've learned a lot from those on road miles and I feel more comfortable when we go fully autonomous, that we have those those miles.

 
Finch Fulton 

Well, we we appreciate the engagement. We've had On the types of operations we're developing, but can you help us separate the myth from what way most actually doing versus what's in the press versus probably what my parents think is going on? which they're all very different? Can you help outline that and clarify that?

 
Debbie Hersman 

Yes. Well, as we all were just home for the holidays, trying to explain to our family members what we do for a living, I can tell you one of the really cool things is when we think about how we educate and how we engage and how we deliver value for people, I think those are really important things. We have to understand that people are different places and their knowledge and with their skepticism to about autonomous vehicles. And so we've got to educate them, we've got to explain what we're doing. We've got to explain why we think that our testing programs are rigorous enough to have the vehicles out on the road. We've got to engage them where they are. And I think certainly at the federal level, at the state level at the local level, we've got to be talking to policy makers and decision makers and bringing them also along on the journey and making sure that we're living up to their expectations. When we talk about delivering value, our early writer program allows us to actually introduce self driving vehicles to the public. And I think this is really important. You've got to engage with the public and help them to understand what it is and get them excited about it. My in lives came out to visit us blast ball. And I will say they didn't dedicate very much space in their holiday letter to me, for the grandchildren. But writing in the waymo got a whole paragraph of how excited they were about this technology. And then it was coming. And so I really do think probably everyone else in the panel can speak to this to experience is really the best teacher for people to understand what this technology can

 
Finch Fulton 

do. Let me come back to that point in a minute. But I want to hear a little bit more about continental and what what role you play in the safety of the systems as they develop.

 
Ralph Lauxmann 

Now have we come more from a traditional ronia because we engage always safety in the industry. When it comes to critical situation, we have the ABS systems in the car, we developed the automatic emergency braking systems and, and based on that safety culture that we already already have introduced based on our technology and our components in the cars, we now do that tremendous step two go in a cooperative mode with all that systems in the cars and that's the huge challenge that we now have to tackle and to master because now we have an engagement of the driver and the technology at the same moment. It's not a helping function when it's getting critical. Now it comes to the point to increase the safety by avoiding that situations and that's the one of the major tasks that we are working on to to increase that road safety while we hand over the driving tasks to the machine.

 
Finch Fulton 

So and so we all have different approaches secular peloton. Tell us a little bit more about what you're doing and how you utilize different aspects to promote higher levels of safety.

 
Josh Switkes 

Yeah, so a key aspect for us, as I mentioned earlier is vehicle to vehicle communication or connectivity directly between vehicles. And it has a profound advantage over any sensor. So, you know, there's tremendously, you know, powerful sensors being developed constantly, you know, out on the show floor, you can't, kind of can't turn around without bumping into another LIDAR company or radar company. And that's great, the sensors are getting better and better. But what vehicle to vehicle communication gives you is direct knowledge of the actions and the intent of the other vehicle. So any sensor all you can measure is the motion of the vehicle, right? You can measure position, relative velocity, you know, potentially other other aspects of motion. You can't get the action or the intent until that motion has occurred. In heavy trucks. This is especially powerful because when you know when the brakes are applied navvy truck there's a significant time delay a fraction of a second but that's that's a big deal when you were talking about safety to me Could time delay between the application of the brakes and the actual slowing down of the truck. So even the world's best sensor you're going to be detecting when that truck is slowing down with VTV we don't have to wait for that. So right when the brakes are applied on the lead vehicle in a in a truck platoon on our following vehicle, which is automatically falling behind that lead truck, we can react before that lead truck has physically slow down we're applying the brakes and that following truck so that's a key aspect of safety for us. Of course with that comes you know all the safety engineering of what if you lose that communication, you know, how do you ensure its secure from a cyber security standpoint all that is super important. But we enable the say we think of it as enabling the safety through the VTV and combination of VTV with sensors and then ensuring the safety through the proper design and security and so on and proper validation and testing. touch on one other Eric safety as well because it's sort of a neriah I'm very passionate about which is, I think over the past years as automated vehicles have gained, you know, gain notoriety and popularity, there have been a lot of claims about how much safety is improved by one system or another. And there has been far too little discussion about what the benchmark is. So we've been very careful when we design our system, we said we want it to be our goal is to have it be significantly safer than then the trucks on the road today, but specifically, significantly safer than the safest trucks on the road today. And the reason for that is, if you want it to be safer than the average truck on the road today, you could just go buy a truck at the dealership and you're going to be a lot safer than the average truck because a lot of them are older and have older brakes and so on. So we said that benchmark needs to be the safest trucks with modern equipment, modern safety systems, etc. And in our engineering work was all focused on metrics that guided us towards an improvement from that already very high benchmark.

 
Finch Fulton 

So you highlight something that we should always make sure that we include in the conversation. It's when we talk about safety, the reason why safety is so important. You may know and then you may have heard these stats before, but they're sobering. 36,500 people died on American roadways last year. 36,500 people died on American roadways last year. Now human error plays a significant portion of that 94% of these fatalities can be attributed to human error being a part of it, whether it's drunk driving, distracted, driving, drowsy driving, whether it's speeding, or even not wearing your seatbelt. Now, automated vehicle technologies are not the silver bullet to fix all these things. But even with some of the vehicle functions that are available today, automatic emergency braking, lane centering, things like that nature, those can have profound effects on the safety of our driving going forward, even if we want to innovate further. But one of the other things that you mentioned, if we talk about safety, we talked about the efficiency of the system. Both of those things touch on the DAX and you may have seen some of the news About the 5.9 gigahertz band a spectrum This is the band that these vehicle functions operate in. The FCC is looking at a reallocation of this band of the 75 megahertz in this band, a significant portion of it to be used for other purposes that aren't related to safety. Now, this is a problem that department transportation recently published a study that showed based on the interference of unlicensed Wi Fi near this safety band, and we're talking about taking the majority of the band, it's all gone. All it will not work. So we have problems. We publish them, we've been talking about them. And I know you all have concerns with this because there's been billions of dollars invested, but also the safety outcomes that can come from vehicles knowing what other vehicles intentions are. Those are critical. And so if you're paying attention, they will be federal docket that interested parties should be engaging in, you can go to the Department of Transportation website, at a safety spectrum page on it, you can find it. If you Google, it actually don't have the top My head this specific but transportation.gov backslash safety spectrum, something like that, you can find a lot of these stats. And it's important to be informed and engaged in this because this is critical and it wreck it makes up an existential threat to the future of connected vehicle technologies and all that they can bring. So this is something that we should all be paying attention to.

 
Josh Switkes 

Maybe I can add to that, because obviously, I'm very excited about VTV. The way I think about it is if we use that spectrum to allow people to watch more YouTube videos, people will watch more videos. I don't think last year any people died due to a shortage of YouTube videos on their cell phone. But 36,500 people died in vehicle accidents. So to me, it's an obvious choice of use that spectrum for safety. The benefits of it will increase over time as more technologies utilize it and more vehicles. But But even if you saved one life with that, that's more than we're going to stay with with YouTube videos with that spectrum.

 
Finch Fulton 

Do any of you other I know you also talked about some of the connectivity aspects, he tell me a little bit more about what Continentals doing.

 
Ralph Lauxmann 

Now, so of course, we believe we believe in the technology, because we see that from the view angle of a car, you cannot see everything. Now, if you only have the sensors that are mounted in the cars, you cannot detect the environment in the full spectrum. So you need something out of the car, and you need the communication between the car and the outside intelligent infrastructure. And therefore, you need that communication and low latencies times also to have the signals at the right time available so that the car can react on the information that he received that it receives from the outside. And so we work on technology to solve, of course, the communication aspects. And we are prepared now to bring something into the market as soon as we get the guidance from the government. It's prepared and as soon as the guidance is there as soon as we can bring it in. The market.

 
Finch Fulton 

Absolutely. So let's talk a little bit more about partnerships and how we engage at the state, local and federal level on these technologies. I know Aurora has talked a lot about the partnerships and waymo as well. I know from Department of Transportation, his point of view, we very much want to work with all of our stakeholders to bring in the types of operations that the local population feels comfortable with. So we've started we provided $60 million in automated driving system demonstration grants that were awarded, then the last year for areas that not only can prove out safety cases for the best ways to use and engage with these vehicles, so that people can understand what they do and what they do not do and what they're being developed to do in the future. To also generate the data that we need to update all the rulemakings that we have coming forth, to set new standards for automated vehicle technologies and the types of operations that trucking companies, for example, need to use we have 10, coming up from nitsa, and at least one major one from FMCSA that are on the fall agenda publicly available will work with these things. But the idea that we need to bring in these operations and to learn from the people and the operations in the communities that want to be engaged. Tell us a little bit more about that and how that's affected what you've done. Either one.

 
Sterling Anderson 

you can go ahead. Sure. We've. So we see, first of all, deities involvement in this in particular kind of the latest guidelines has been enormously helpful. We think this is the you talk about partnerships and kind of ecosystems that are needed. This is a wide ranging problem, right. And we view when I when I say safely, quickly and broadly as our mission statement, each of those words is intentional, and their orders intentional, right. And if I made the order of your questions appears to be kind of be following the the kind of appropriate rhythm so given that I'll take kind of the quickly in the broadly bit, we think we bring a safe product to market most quickly. When we work in collaboration with those who are experts in their respective fields. We built a rora from a uniquely concentrated set of experts in self driving, we will build a safe driver, we will not build a car. And and nor would we be very qualified to do so. But we have an awful lot of of tremendous talent in the world who can do that very, very well. So we took the approach of we will build a platform for the road driver that will bring that will incorporate and allow us to play our position and others to play there is this is why early on, you saw us putting in place the pieces that were kind of the long poles, right partnerships with various automakers who would build the platforms that we would establish a Functional Safety case around now to partnerships with carriers. Those ultra who will ultimately deploy this product, you'll see more partnership analysis from us around fleet management service providers around others who to have to take part in some of this. Then on the government side, you've also seen you know we've we've collaborated pretty extensively with various state and and federal agencies. We were I think the first to receive automated here Testing permit and state of Pennsylvania mentors. From there. We view this is a, we will not bring a safe product quickly to market and ultimately scale it broadly without collaboration between many of these different groups. And we think that's important and to kind of bring the public and others along. It's enormously important that you show your work in this business. Right. Many are understandably concerned that when we go to commercialize something, that there are major liability problems, right, the first accident there will be an accident by an automated vehicle. There have been and once we commercialize, there will be no matter how safe we are. Not no matter how, sir, caveat, provider, we're not perfectly safe, which we probably will not be. When that happens. The receipt of the public of that news, that receipt of regulators of that news, that receipt of the rest of the industry, that news is going to be fully a function of how informed they are as to the data that made us believe we were objectively safer than humans when we deployed. And so the part of the process for us is Working with federal, state and local authorities to ensure that kind of the data that is still data that helps us establish a Functional Safety argument for before the system is available. So that these accidents, these incidents can be taken in stride. And so, so we think that's an enormously important part of that collaboration that, for me, personally, is extended for many, many years with nitsa. And with other men. Lesson inspiration, and and we hope will continue kind of in as we go forward. Absolutely.

 
Debbie Hersman 

So, I'm sorry. Go ahead. So I'd say that the partnership is a issue is really important. And obviously, federal, state and local partnerships are key and critical because all of those parties have a role and have responsibility and what goes on. Let me just give you a couple of examples of some of the partnerships that we have focused on it waymo, and I don't want to repeat some of the things that Sterling said so I'll focus on Some of the other things that we've worked on, one would be our let's talk self driving campaign. And we really focused in Chandler Arizona in the Phoenix area on educating people with partners like mad and AAA about why self driving and what self driving is. And some of this is about educating and providing information, setting up websites giving people access to information being in the community. And so it's about having those engagements and being there on the ground with local community members. The second example that I would share is our partnership that we had with first responders, law enforcement firefighters, we actually published a guide of how to address self driving vehicles, specifically the way mo vehicles in the event of a of an incident. And so we also have offered training to local first responders about how to do that showing them our vehicle explaining to them How to engage with the vehicle. Again, this is published on our website, it's available for anyone. We've also produced videos, training videos for, for first responders that are free of charge that anyone can access. And so I think all of us recognize the incredible responsibility that we have in this nascent industry to make things safe. And we all need to play our part. And I think as industry leaders, it's incumbent upon us to help make that progress, step by step, transparent, you know, when it merits sharing, and I think working together, inappropriate venues. And I really appreciate the guidance that you all have released, not just 4.0, but two and three before that, that really encourages the industry to continue to innovate. And I think what you shared today is really trying to say, our place in the world and where we want to be with respect to making sure that we can innovate, but do it safely. responsibly.

 
Finch Fulton 

So you highlight something that I think we need to make sure we're very clear on. So when we talk about the safety of a human driver and improving that, that's not enough. Not only is it not enough, and we shouldn't accept it, but no one will accept it. We see what happened when there was a crash involving the testing of an automated vehicle that struck and killed a woman, and what that has meant for everyone and the refocusing on safety from all of the environment and all the community, we cannot take this approach seriously enough. And so as we work together in partnership with sharing of information, sharing of practices, publishing, voluntary self safety assessments, ensuring the entire communities and formed, we need to take more opportunities, and you'll see more announcements and news from the Department of Transportation about our role in this coming forth in the weeks and months to come. But we need to make sure that we're getting that information out there not only in learning from each other before crashes happen, but also trying to make sure that people understand what is being deployed, what technology it is, and how it impacts the operations and the roles of the human and how it works the environment around it. It's also very important that we are clear in the terminology we use. And it's something that frankly, the Department of Transportation has to do a better job on. And we're working with a number of partners, we've been proud to engage on a clarification of terminology working with pays with triple A with Consumer Reports, sa, we're trying to make sure that not only are we clear about what it means for the sophisticated audience that loves talking about level one through level five of automation, and what it really means that's not going to work with my parents, they don't know what that means. It's good enough to say it's fancy cruise control, and you don't you can't afford it yet. But eventually, we're going to have to do better at this. So we do have to work in partnership to make sure we're clear about what a vehicle can do, and cannot do. And as we introduce new technologies, and we're developing new technologies that improve levels of automation going forward, we can make sure that the an engaged public that doesn't want to have to be a wonk understands what their roles are and how they can rely on this technology. So I don't know if you want to talk a little bit about continental and what you do and how you make sure that your partners are engaged and understanding your technique.

 
Ralph Lauxmann 

First of all, I fully support that we have to be clear in the wording because when we talk about autonomic functions, then we can put a lot of different use cases under that wording. And that's totally misleading. So if we would find a way to define that words a little bit clearer, and put use cases under that, then it must be that I think is the right way to, to, to get the rights transparency in all that functions that the car and the infrastructure is then realizing, and helping all the guys who are then driving in their cars, to understand what's going on around them and what they can believe in, and where are the limits of the functions. So the wording is one of the major tasks that we have to solve in the next two years in a minimum. So how we work together with the with with our, our partners, that's not so different from the past, so in a role as a tier one supplier so you will see supplies, the components to software to the car manufacturers. And you develop the things to a certain level to come together with the car manufacturers, and then it goes in serious production, and then the cars gonna be sold, and you get experienced back and the next generation of the components is doing the job will be better than the last generation. And I expect that that situation will not change in the future. Even when we go in that automation mode. We as Sterling said, we're not going to be perfect from the beginning on, but no car has been perfect from the beginning. But we are safer than the less than the level that we can reach with today's functions in the cars. And I think that's the main target what we should reach

 
Finch Fulton 

when we are saying higher levels of safety playing out in the higher levels, the improvement of the safety numbers overall. But so let's look at a couple major category of writer of driver and of professional certified truck drivers and how they engage with these technologies and how we need to be thinking about what their world looks Looks like over time, because obviously we assume something's different if somebody's sitting in the back of a self driving vehicle, or if they are professionally trained and certified driver that knows what they're doing, and they make a living off of it. So tell us a little bit more about that. And then engagement and making sure that over time we do a better job and and always improving on what the vehicle does and does not do. And what the expectations of the passenger driver a professional driver are. Whether it's automated vehicle technologies incorporated or an activity or

 
Sterling Anderson 

Okay, you're looking at me, so I'm assuming I invite the whole panel. Let me say a couple of quick things in one kind of ties in your last question. It is imperative. I think the people on the stage as well as others who are developing this technology. If the public cannot trust us, because we say overly aggressive marketing, we use overly aggressive marketing language, or we get ahead of ourselves in terms of kind of promises to the market. This will tank the industry. I think you know, there's been a lot of talk of, you know, are we in the trough of disillusionment? Are people really kind of coming to this realization? We are and you know, why not because the technology is delayed but because the industry generally and the kind of hype machine that built up around it got a little ahead of itself. And and the understandable pullback is the kind of general public who listens to the breathless, kind of, you know, announcements of various people finally starts to realize, Oh, wait, there's a gap here between kind of reality and what's been kind of sold to us. So it is critical that we are open and honest with the public as to both how we develop what the statistics are that we're using to ensure that what we're doing is safe, but also kind of reasonable timelines for for kind of what we're putting out in the world. I was just at a dinner the other night. We're in another attendee piped up and said, Yeah, you know, a lot of my lot of my employees are, you know, they ride their self driving car to work these days, and so they just watch YouTube videos on the way And, predictably, she mentioned you know what car most of them drove. And that's kind of the hair on the back of my neck Rose is I have some history kind of with that and the realization that no matter how many times I or someone else will say it, these are not self driving systems. They will not they today and and for people to treat them that way is very dangerous. There's also the and that ties into your next question, right? How do you deal with professional drivers? In my view, and in my experience, as Josh mentioned, it goes back 15 years kind of in the space. In my view, it is extremely difficult no matter how train they are, and I'll use for you the proxy of the airline industry. Several of the crashes that have happened in recent years, largely is as it as a result of, you know, various instrumentation failing, pilots believing they're in a stall or you know, airspeed indicator goes wrong, and making the wrong move. Those pilots are trained for thought thousands of hours, thousands of hours. That is that is a pretty high level of training. And yet mode confusion is a problem in that space whenever you expect a human to take over control in a fast paced or risky situation, particularly is the frequency of those situations to clients, the system gets better. It gets worse, not better. And so at Aurora, we took the view of we are developing a self driving system that does not have the expectation of a human fallback neither in the car ultimately, nor behind the scenes and until the operation center and we published recently kind of our position on this was in reference to our teleo system system that we're developing. So a system system establishes the expectation that the car itself has full responsibility for maintaining safety on the road. There is not a there's not an expectation that there's there is a human anywhere in the car or elsewhere, who will arrest a situation before it becomes safety critical. The car must be designed such that brings itself to a safe state. Now from they're applying higher level reasoning to ensure higher levels of service and so that the vehicles, you know, should they be confused aren't blocking traffic for long periods of time. That's where for us things like tolia systems come into play. And so, in my view, driver assistance, once it surpasses a certain level in a certain kind of establishes a certain level of reliance that is detrimental to the safety of the roadways.

 
Finch Fulton 

Do any of our panelists want to weigh in on

 
Josh Switkes 

Yeah, maybe I can I can mirror add to some what Sterling was saying about mode confusion, and sort of trust in the system. So when I started my career in autonomous vehicles, there weren't any automated vehicles and it was like 2000. Well, there were only you know, prototype and demo vehicles. But I was working at Volkswagen with the Human Factors department looking at how to how to drivers detect a problem with their automated vehicle. So we didn't have real out of a vehicle. We have a vehicle with a one way mirror behind the front seats. Someone was sitting back drive a car. That was that was me because I didn't speak German. And my German colleague was sitting the front with them and you know, help the the test participants were looking at, you know, I would drive for a little bit, I would shut the system off, we'd see how they detected that problem. Was it from crossing a lane line? Was it you know, how did that How did that happen? The more interesting than the official result of it was I got to watch all these people as they gained trust in the system. And keep in mind, they didn't know anything about the details of the system. They didn't even know it wasn't the system. And it was it was an American sitting in the backseat. But they gained trust extremely quickly, because the system did what they expected the first time. Right, and we turned it on, I would drive I could drive pretty naturally. And they would trust and immediately they take their hands off the wheel, they talk to my colleague in the passenger seat. So that was both encouraging from kind of a marketing standpoint of Okay, people will trust the system very quickly. But it was also I found that very cautionary experience, because it means people will trust it. Just because does the right thing the first one or two times and that's that's why we you know, these problems result with Tesla autopilot, right you're, you're driving it It drives itself you like great it drives itself I'm gonna watch my YouTube videos and you you know you may or may not know the details behind it and know that you shouldn't be doing that but it seems like it's a in levels automation seems like it's a level four system Some people use it as a level four system so i think you know with that being said, like Sterling was saying experienced drivers we think of them as a resource for designing our system, but we don't rely on their experience in their use of the system. Instead we tried it our system involves a driver you know today in both trucks with the next system, you know, we'll have a human driven truck and then no no driver in the following truck. Either way, there's a human involved in our system on the in contrast to Aurora and because of that we put a lot of effort into what does the driver see what do they hear what do they feel Sterling mentioned mode confusion. That's extremely important to us for the driver to know what mode the systems in is it active? Is it deactivated? What is it capable of in this current mode, so the driver can know how to interact with it. So that's the approach we've taken.

 
Ralph Lauxmann 

That's what I really support both both statements. And we spend a lot of time in developing a kind of communication between the automated driving car and the passengers as well as the pedestrians. So, such a car has to communicate with the outside how even when, when pedestrians want to cross the road young you normally go with an eye contact with the driver and and try to realize if if he has recognized you, and if that happens with a car, well, no driver is inside. You need a kind of communication by lights or sound or whatever, to gain the right trust and to believe in that the car is recognizing you and that you can cross the street and so we developed monitors show signs to the outside and also to the inside so that the passengers in the car believe in the technology that the technology is capable to detect outside in the right way. And that it stops if an obstacle if an obstacle is on the road. And so, yeah, so that that trust level, I think that's one of the key, the key of future development targets that we have to realize, to to make the introduction of that automation level successfully. So let's talk.

 
Finch Fulton 

So let's talk about the biggest challenges that you see coming forth. I know, one in particular is what are the biggest challenge you see in the world of automated connected trucking that you see coming forward that will have to be overcome?

 
Josh Switkes 

Sure. So Well, normally, we haven't spent, you know, half an hour about safety. We talk a lot about safety, but maybe some specifics on the safety side. We see that the have safe automatic connection. Trucks means you have to have the right functionality. And you have to have lamented that and validated and tested that that right. So and sets off, I think in the kind of automation world that got lost a little bit in the early heyday five years ago, where everyone was excited about sort of computer vision applications for automated vehicles, and forgot that the whole system has to be safe together. So specifically on a heavy truck, it means you have to have the right actuators. Today's trucks on the road don't have redundancy in their braking, they don't have redundancy and some of the data pathways and things and they don't have the right steering actuation to to be an automated truck. But that's all in process, you know, suppliers like continental and their colleagues and competitors are developing all those things. And that's coming. Then, you know, the second thing I'd say is, people often talk about the highway environment and trucking as really attractive for automation because it's much simpler than urban complex environments. That's true in some ways. But what's often left out of that analysis of those statements is the braking distance of trucks on the highway. So at high speed heavy vehicle are much, much longer than a passenger car at low speed. So if I'm in Mountain View, and there's a waymo vehicle, and it stops, you know, incorrectly going 25 miles an hour, not a big deal it which wouldn't have which would never happen. But if it maybe not way most of you know zooks vehicle and safe is on the road. If that happened, not a big deal, because false, you know, false positive at low speed, not not nearly as big of a deal, as if you're going 70 miles an hour and a heavy truck on the highway. So heavy trucks on the highway have major challenges. How do you get down to a low rate of false negatives, meaning there's a stopped car 800 feet ahead, you better be slamming on your brakes, but also a low rate of false positives, meaning if it's not a stopped car, but it's something that looks like a stopped car, you're not slamming on the brakes, because every time you That you're creating creating a safety risk in that heartbreaking stop. So I think that's sort of the biggest overarching challenge around the highway autonomous trucks is getting to that long range of sensing and perception with a low rate of false positives and a low rate of false negatives. And that's more than just the sensor challenge. That's a sensor and software challenge. we simplify that tremendously by following another truck. And that way all we have to worry mostly all we have to worry about is what's between us and that truck in front of us. So much shorter range of perception challenge. If I could just add to that,

 
Sterling Anderson 

agree with everything you just said. There are a couple of other things that I find or I think less well understood to the general public or even that some people fall with space. As I hear frequently the same thing as must be easier to drive on highway right. It is, but there are there are a couple of things that we want you to know first Josh Manson's kind of the the difference in severity right? Typically we think about the product of exposure and severity, how frequently does something happen? And how severe is it if we make the wrong decision, when that thing happens, frequency is much lower on highway severity is much, much higher. At the same time, we believe the kind of ethical standard for deployment of any of these systems is better than human. And guess what humans are also better on Highway to the tune of four x or so fewer collisions per mile than you would experience in humans experience, kind of a more congested urban environments. And so when we think about that, we think look, in the kind of edge cases that you're going to encounter on a highway, maybe look just as weird, right? You'll have a guy changing his tire and running out across the road, something else happening. And so we view this as you know, at Aurora we started over the last few years, we focus specifically on very dense urban environments, not because we didn't care about highways, but because the efficiency of that development is much much higher there. We encounter far more incidents per mile of interesting events. Now as we as we move forward, you know, one of the other things that we thought safe self driving on highways was really missing. It's just as he also mentions a compelling sensing solution for long range sensing, right, many of the pulse modulator ladder units in existence today, and we've been searching kind of the world of LIDAR for several years now. simply will not get you the kind of range that you need or the kind of data that you need. We were really excited this year last year, I guess now to have acquired Blackmore company that for 15 years, has been really pioneering frequency modulated continuous wave LIDAR, which the difference in this is you think about sort of conventional LIDAR conventional, most most lighter that exists today in the automotive world, effectively acts as an amplitude modulated unit, it sends out a pulse of light into the world. And it looks for kind of light returning that exceeds some threshold and says, Okay, what was the time of flight and therefore, what is the distance and here's a point in space, where we think it would from which we think it reflected. The problem with that is particularly to get to long ranges. You have a power problem when it when it as it relates to safety. But be you don't know if that point in space was hitting st post, whether it was hitting hitting a rock the roadway or something else entirely. Now when you have a frequency module, a continuous wave system, you have instantaneous relative velocity of that object. And now you know, that object that I've just seen on the highway from my truck at 300 meters out is traveling at 70 miles an hour, that's not a pedestrian, that's not a rock, that's not the ground, that's a car. And that that really opens up a lot of opportunity for safe operation at high speeds. And so for us, this has been about a transition from urban operation to kind of, you know, hub to hub logistics sites applications, really with the unlock that is this this key kind of technological enabler, but as Josh says, the you know, I would I would encourage anyone who really is curious what's really more complex to really study both the right incidents that humans tend to encounter and these but also kind of the severity of the incidents. And I think you'll find that it's much, much more similar at the end of the day between urban and kind of rural highway settings and then you might otherwise have assumed.

 
Finch Fulton 

So, with the time we have left, it's probably a good idea to ask the question, there's the name of the panel. What's next for vehicle automation? And I want to be thinking about the next three years the next five years What should we be paying attention to from your company and from the ecosystem as a whole and I don't know who wants to start Ralph I think I saw you wave your hand

 
Ralph Lauxmann 

Yeah, so so I'm I'm I see upcoming in the next step, that we will realize that valet parking function in the cars in the standard pedestrian cars based on the sensors in the cars may be based on sensors that are mounted in the parking garages or or wherever and the parking lots, so that that kind of parking maneuver littering is is automated I think that's going to be the first step and that's really a cool thing because even in in in the US you have sometimes the problems to find the parking spots in Europe it's much worse and in Asia it's horrible situation if you are in Shanghai or or in one of the super mega cities, you have to drive around up to 30 minutes to find the proper parking spot and that's that's

 
Debbie Hersman 

take away my you don't have to have a parking spot.

 
Ralph Lauxmann 

Yeah, that's right. But they're still not existing in Shanghai. If you take a taxi, you do not have to problem but everyone is using his private Leone car area, and they go with the privately owned car in that areas. And so it's a it's a, it's a for the for the city and for the people super function which helps them to save time, and also reduce the pollution in the cities. The second step that I see and where we are, I am Really focusing on also personally is that we bring the people movers on the road and so that we have the possibility to have that new kind of mobility scenarios in the cities where you do not have to wait at the bus station for for a bus who's doing the predefined routing, you have your mobile phone and you can order a people mover and tell him tell the car that he that it brings you in front of your door and you go that last meters last mile based on your own demand and your your car that that brings your safely to your door and at home. I think that would be very exciting.

 
Debbie Hersman 

I'd say for waymo. The next few years are really about rolling out the technology responsibly to eliminate Those common human causes of crashes and to improve the safety of the roadways that we operate on. And say We do this through continuously developing and expanding our product, but also looking to successfully leverage what we're doing in our rideshare applications, to trucking to delivery and potentially to personal car ownership. And obviously, some of these are longer term activities, but I think those are those are things that are going to make sense going forward.

 
Josh Switkes 

Yeah, for peloton, so we've been piloting our platooning system, the level one system where there's a driver in each truck, and they're automatically following each other and piloting it for the past year or so. With a number of customer fleet customers, showing great fuel savings and showing what appears to be a great safety improvement I say appears because when you set the benchmark very, you know, a very strict benchmark like we do. It's a collision every, you know, many, many millions of miles and we don't have those miles yet, right? So statistically, we can't say we're showing it but but the future is we expand those and ardra deployments, you can see more and more trucks output turning, and then you'll start to see our automated following trucks in pilot phase with on the highway with, you know, with safety drivers. And I've also had that sort of progression I just described as I think what we're going to say, across the board and autonomy, meaning you know, for years people said What year is there going, you know, are we going to have an autonomous vehicle and you know, that year shifted later and things you know, if we were if it was a binary switch of having an autonomous vehicle somewhere, it's already happened right waymo Volvo safety drivers out in Chandler, so that happened. But along the way, what people realize is it's not flipping a light switch, right? You don't just suddenly have an autonomous vehicle that can be everywhere in the world. You have it in Chandler and you have it on passenger vehicles and low speed and for us will Have it as an automated follower and so on. And then you branch out from there. So I think people are realizing more and more, it's going to be this gradual progression of specific use cases and operating environments. And then eventually you get to the full, the full, you know, almost everywhere type of automation, but that's going to take a long time.

 
Sterling Anderson 

last three years for Roro have really been about building the foundation, foundation of our platform requirements that allow us to be platform agnostic and operate in anything from a class a truck to a small passenger sedan, laying the groundwork, the foundation for our software architecture, building the beginnings of Functional Safety, architecture, building starting to build partnerships across the industry. 2020 is really going to be a year of framing a lot of that together. We think the first application in which will likely deploy a product isn't the movement of goods and we're very excited About the partnerships that we're establishing this space, about this technology we've brought under our roof to allow that you'll see a convergence of many of the investments we've been making in, you know, various spaces in, start to be embodied, kind of end up in a product that ultimately will roll out. So the next, the next couple of years will really be about us, refining that product, ultimately removing the safety driver and deploying it.

 
Finch Fulton 

So answering my own question, one of the things that I'm interested in that I think will connect to the whole picture for vehicle automation, but it's really simple, and it's data, right, so how can we better utilize data? How can we recognize the value of the data generated by the vehicles? And how can we figure out what are the outcomes we're trying to solve for right? So we have very legitimate and understandable safety outcomes that we're all working towards, right. And so how can we figure out what specifically do we need to know? How do we get that information How to regenerate it. How do you protect it along the way, because if we, if we identify a use case, an example uses the work zone data exchanges, you know, when work zones go up, they're required to flag where they are. And there's not really great systems for highlighting in maps around the country where these are, I mean, a lot of these states and the people that have to flag these use fax machines, and I don't know how well that works in with your ways applications, but that that doesn't cut it right. So we have a use case. And now we have willing partners that are helping us to figure out how do we create data? And how do we provide that data so that we can all incorporate it into our systems, whether they're connected and automated vehicles and whether or not they're just basic data applications that we can use. And this also looks into if we look at the connected vehicle data that is generated. We see in our Tampa site, for example, we have connected vehicle pilot projects that we've been deployed, they're out there, they're successful. They're providing data and real time back the United States Department of Transportation, and we're talking about operations in Wyoming, in New York City and in Tampa, using different types of connected vehicle technologies. But suddenly, we're learning about systems global operations, right, we're able to tie together certain things that we didn't know before and figure out better ways to use the system. And this is important because as we look at these little bits of data and providing them to cities and provide them under users, you're starting to see use case applications out there today with the people that are doing it well, providing this data to first responders, for example, so they can be better positioned and anticipating crashes. We're seeing lives being saved today by reducing the reaction time needed by first responders because they're already in the areas where we know that there are problems, we can already identify safety problems and preposition or make changes in the system so that what's causing the safety problem can be addressed. So one of the things we're doing is using some of these data packs using the works own data exchange to get agreement on types of data types of uses, build out that trust so that ultimately as we evolve higher levels of technology and sophistication, we have a platform to build from because suddenly we have a way that makes sense for all active parties to get data exchanges for automated vehicles, right. So this is something that ties together work that's going on today, that can be used to save lives today. They can lead to this greater future that everybody wants to do but recognizes the value of the data and the people that generate it. So I'm excited about that. It's about safety. It's great. I will use this opportunity to also highlight again, one of the things but we essentially have three requests that I'm going to be asking from you after this panel. First of all, this is just a mere request, you should read this document. It's wonderful at 4.0. Again, it highlights the 38 federal agencies that are engaged and involved in automated vehicle development. This is important because there are billions with a be billions of dollars in federal funding going towards basic research. And these activities are billions of dollars in operational deployment grants that are available to engage stakeholders today, they can utilize this to help support the better development of these activities. We have dozens of activities going on, where we need to engage with interested in engaged public so that we can develop these together. There are many things that people just don't know that the federal government's doing. My favorite example is the United States Postal Service. I own 255 Vehicles, they put out an RFP for automated vehicle technologies. And they've shown if you can automate even at lower levels, low speed routine maps, delivering to every house six days a week. Well, that's interesting. And if they can save, you know, one of the benefits is they can save seven seconds per mailbox. As you roll up, put melon, hit the button, while it's going, you're gathering the mail, get to the next mailbox, hit it again, go. But imagine what utilizing that a government funded initiative utilizing that to figure out if we want a world where we have all these technologies that engage How can we use these sort of opportunities. So the first request I have is read this document, it is a it lays out the map of everything going on in and as a tool for American innovators to figure out what's out there and how they can tap into it. So we can all work together towards the type of transportation system we want in the future. Read it and then we're also going to ask for you to make comment to a public comment. We put everything out for public comment, so that you can not only tell us, you not only figure out what makes sense for you, but you can tell us the dots that we need to be connecting better. How can we drive together the third DHS, federal agencies, what use case makes sense for you that we're not doing so that we can bring all these things together to again, improve the outcomes that we're going to see. So that's the first request, read it, love it, or hate it. And provide public comment because our next step will be laying out a comprehensive plan from where we are today. All the regulatory updates were happening, the involvement and safety standards development over time, but what does it take to get from where we are today to the safe and full integration of automated vehicles into our national transportation system? That's our next big task. So we need your comments because we're shaping our work now. Alright, my second ask is to the US do t has a booth 1501. It's right out here. If you go around the corner, you'll see it's right in the entryway. We have lots of fascinating stuff going on there. You should come and check it out. We have our university transportation centers that are presenting they're showing a lot of the innovative work going on in transportation. It's wonderful stuff, please come check it out and talk to the very smart people we have staffing that booth in my last and final request is to just join me in thanking our members on the panel for their participation here today and thank you so much for being here as well.

CTATECH-PROD3