We didn't hear anything about the rumored partnership between Google and Ford at CES this week, but Ford had lots of other news. Two of the company's biggest announcements was that it would be tripling its test fleet of autonomous Fusions to 30, and that Toyota and two software companies would also be adopting its SmartDeviceLink (SDL).
SDL is an open-source version of Ford's AppLink, which allows developers to make apps for Sync, Ford's infotainment system. By making it open-source, Ford is attempting to make SDL an industry standard for carmakers' embedded infotainment systems. Along with allowing car companies access to apps already built to work with Ford's AppLink, adopting SDL would allow developers to only have to design apps once, rather than different versions for different systems. Because its open-source, another advantage is that anyone using SDL can contribute to its betterment. Other carmakers looking into adopting SDL include PSA Peugeot Citroën, Honda, Subaru, and Mazda.
Both of these initiatives along with some fall in line with Ford's push to expand its business and influence beyond just being a carmaker. To find out more about this news and what it means for Ford, R&T sat down with Raj Nair, Ford's vice president of product development and chief technology officer. The following is a slightly edited version of our conversation.
* * *
R&T: Could you talk a bit about your expanded fleet of autonomous Fusions and what exactly the next steps are for your autonomous-vehicle program and development?
Nair: So it's our third generation of autonomous vehicles. Our second generation was the Fusion Hybrids we have on the road right now, and our first generation were the original Super Duty trucks we had in the DARPA challenge. This generation, certainly we'll focus on the algorithms, as well as the sensor componentry. On the sensor componentry side, this third generation will be using Velodyne's third generation LiDAR technology, , which is a more compact, longer range, more precise LiDAR system, which will help us on our 3D mapping and 3D sensing. And then on the algorithms, it's just a continual refinement of our algorithms against all those real-world scenarios that you continuously find and have to adapt to. The end goal, in our view, in a four-year time frame, somebody comes out with a vehicle capable of SAE Level 4 autonomy. And the constraints that we see in Level 4 are a geo-fenced area that's 3D-mapped and the weather conditions that are going to be conducive to the sensor suites that we'll be using.
(Editors' note: There are two types of level designations for vehicle autonomy. The National Highway and Traffic Safety Administration (NHTSA) uses a four-level system while the Society of Automotive Engineers (SAE) to designate different forms of autonomy. For SAE, the Level 3 systems we are now seeing such as Tesla's Autopilot enable a car to drive itself in certain situations while still requiring human intervention. Level 4 is when a car can take care of the majority of driving tasks without driver intervention. Level 5 is true autonomy.)
R&T: When it comes to LiDAR specifically, Elon Musk has said you don't need it for autonomous cars, and though they are getting cheaper, the cost is still prohibitive. Do you see the technology getting to a price point where they are more feasible to have in consumer vehicles, or are you just using them to build the maps that you need?
Nair: I can't speak to Elon's experience or Tesla's experience. Our experience is that we need it on the autonomous vehicle. And it will be using the LiDAR in context and then comparing that to the previous map data that it has to have a better understanding of exactly where it is. And so we believe it is required on every vehicle. Not just ones that are going to be out there mapping.
R&T: With SYNC 3, owners have the ability to connect to their homes over Wi-Fi, so I'm curious if there is going to be some effort to start collecting mass data in the same way General Motors and Volkswagen to start informing your own algorithms and maps.
R&T: There is going to be a tremendous amount of data available once we have the vehicle as a device on the Internet of Things. Even before full autonomy, the aspects of how much vehicle data can be available, whether it's traffic patterns, or congestion, or parking spots, or weather conditions. I think you'll see a trend towards that. Our view is that any data the vehicle generates first is the customer's data, and whoever owns that vehicle owns that data. So we'll always have some sort of opt-in policy. Do they want to opt-in and make that data available? And we'll have to make sure that any personal, identifiable information is still protected and we're just talking about using that data in the aggregate or for a specific service that the customer finds valuable.
R&T: I want to talk a bit about SmartDeviceLink. Toyota announced this week that it officially signed on to the standard, and a few other big carmakers are interested. You haven't said this, but with carmakers able to control who has access to what vehicle data and the push for everyone to adopt one standard, it seems like a really good way to kind of box out Apple and Google from getting too far into the car. Is that a little bit of the strategy?
Nair: Well, I wouldn't say we are boxing them out, because obviously we announced with Sync3 that we are enabling Android Auto and CarPlay. But there's clearly a lot more data available to the car, and that is one reason why Sync is still a pretty valuable alternative to the customer as well. That's why they have the choice. With Ford Sync, we think it's important through AppLink that developers can develop the same types of apps that they would for CarPlay or Android Auto. The big names obviously being Pandora and Spotify, but there are many, many others. And then if we can make that a standard that the industry can use for the industry embedded systems, that's obviously much more efficient for the developers. The fact that there's going to be a lot of vehicle data that is made available to the native system that's not available to Android Auto and to Apple Carplay will obviously enable us some incremental value to the customer that can't be reproduced.
R&T: But it is a conscious decision Ford is making to limit the vehicle data other parties have access to, right? I'm sure Apple and Google would love if you provided them with more of that data for Auto or Carplay. That's a choice you're making.
Nair: It is a choice that we're making. You'd have to ask Apple and Google how they'd feel about it.
R&T: Well, wouldn't it allow them to make more apps?
Nair: Again, you'd have to ask them, but it certainly allows us to make more apps.
R&T: In your mind, what is the path moving forward for the implementation of semiautonomous to potentially fully autonomous vehicles.
Nair: For SAE Level 4, we think someone will probably have a commercially available vehicle in four years. I think SAE Level 5 is a little bit harder to predict. To get to 10 percent autonomous function of your 100 precent driving experience takes a certain amount of capability, to get to 20 percent takes double that, to get to 30 percent takes quadruple that, to get to 50 percent it's eight times that—it's really an exponential scale. And so the aspects of the solutions outside of a 3D-mapped area are still in a research phase for us. Where as SAE Level 4 capability has advanced enough that we would call it an advance engineering project. We're not in a race to be first. You know, we think we're in a pretty good position, but we think that someone, based on the rate that we're seeing, will probably have something in four years at SAE Level 4.
R&T: In terms of map data, which is so valuable to autonomous vehicles, what is your strategy for building a really robust, propriety in-house mapping system.
Nair: Yeah, currently we're doing our own mapping right now. In the areas that we're testing in, we're also testing our ability to generate those high-definition maps. But you know, we're also open to partnerships on that. Clearly there are some very good mapping companies. Whether it's Here or TomTom or whoever, they obviously have plans as well. We really haven't made a decision yet on what our approach is going to be on high-definition maps or 3D mapping . . .
R&T: Google has good maps, don't they?
Nair: I imagine they do, yeah. I hear they've got an app called Google Maps.
R&T: Is there anything to what we're hearing about all that.
Nair: What are you hearing?
R&T: Just that there's a potential partnership between Ford and Google going on.
Nair: I hadn't heard that. You know, I'm kept in a tight, reclusive, hermit-like existence.
R&T: Good to know. OK, so for something like Level 4, but let's even say Level 3, the silence from the government, at least publicly, has been deafening. There is a lot going on right now, and even with Tesla's Autopilot debut, you haven't really heard that much from NHTSA. Behind the scenes, is there a lot of communication between the government and carmakers about what they're doing—what's OK, what's not?
(Editors' note: This week the United States Transportation Secretary Anthony Foxx that the government wasn't planning on creating any regulations regarding semiautonomous or autonomous cars. This interview was conducted before this news was reported.)
Nair: We've had a lot of discussion, and it's been very proactive. I think it's important that we have those discussion as an industry with the regulators so that we're kind of presenting a common point of view on that. That always works better relative to regulations. I think that in the discussions that we've had, particularly with NHTSA and the Department of Transportation, they really see a lot of advantages relative to safety and congestion and that the vast, vast majority of accidents are caused by human error. Assuming that we're going to hold the autonomous vehicle to a very high level of standard, above and beyond a human driver relative to its capability, they see this opportunity as a safety opportunity, a congestion opportunity. They're very pro the opportunity. But we're still at a maturity level that we as an industry need to work with them on what would the proposals be, what should the regulations look like? So I think the fact that it's been quiet is more about where we are at in the conversation, not because it's an environment that isn't supportive of where we're heading.
R&T: The hand off between the car and the driver is often said to be the biggest challenge in introducing semiautonomous features in production vehicles. Has NHTSA given any indication as to regulations regarding hands-off steering and how that should be handled?
Nair: Well, I think that you're at that right now, at the Level 3. And that's a bit of a problem with Level 3 and trying to increment your way from Level 3 to Level 4. As you get to these higher level percentages of automation and higher level percentage of people's time not driving the vehicle, due to natural human nature and natural human psychology, the driver loses attention and loses awareness of what the situation actually is. And then when the vehicle needs to hand over, it's harder and harder to hand over as those percentages go higher. We've seen this a little bit from the aerospace industry.
When we're talking about Level 4, it's under an assumption that you're in that geo-fenced area. And obviously the weather is reasonably OK, which we hope we can predict with some accuracy within a reasonable amount of time. We'll never be in a situation that the vehicle has to hand over, and so that solves that human-machine-interface problem that you just mentioned. That's our point of view. You'd have to talk to other manufactures—there are others who do think you can increment your way. Our view is we'll continue to work on driver-assist technologies, but Level 4 will not be achieved by incrementing your way. You have to start with an assumption that you never have to hand over.
R&T: We've heard about liability this year. Volvo, Mercedes-Benz, and Google have said they would accept full responsibility if an accident happens while a car is in autonomous mode. What is Ford's position on liability?
Nair: I think the liability question is something that we as an industry, the regulatory agencies, and the insurance companies, all need to work together on determining how we are going to manage that. But overall, if we were to offer an autonomous vehicle, we would stand behind the performance of that vehicle just as we stand behind our current products.