The Future of Cars | Dragos Maciuca

Sorry, that was—

OK. Amazon Echo, this one we announced at last CES. And we're working feverishly now to get it out in the market. Basically, it allows Amazon Echo to interface with the car and the other way around. So basically, the car becomes, as I said, just another T in IoT. And actually, I always find it funny that The phone, we always think of the phone as an accessory to the car, but if you talk to the phone makers, they think of the car as an accessory to the phone. So it's this duality basically of who controls what. But the same thing is happening between the car and the house. So you have an Amazon Echo in the house, you have your car, so then from the car you can actually control things in the house, so like opening garage door, turning lights on and off. Checking your shopping list, or the other way around, you can check the status of your vehicle, lock/unlock doors, start the engine, do things like that. So basically it's this duality between the car and the house. This one seems— may seem a bit removed from what we normally do, but looking at connected bicycles is actually an important strategy of what we're doing, and it falls into the area of mobility. Basically, again, in large cities we can't just sell cars. So we need to look at this ability to look into multimodal transportation. Also we found out through this is that by getting the pulse of the city we found out that bicyclists are also the ones who use car sharing the most. So by understanding how this group of people use bicycles we can also understand in the long run how they are going to use car sharing. How do they go from a train to a bicycle to car sharing? How do they actually move through the city in devices like bicycles that before we didn't understand them? So we basically put electronics on those bicycles, we let them through the city, and it allowed us to understand the pulse of the city and how the The city really lives through the motion of these devices. We then took the same technology and we adapted it to motorcycles.

And this is—

Riders for Health is an NGO, a non-governmental organization that delivers medicine in Africa in rural areas where there are no roads or I should say the roads are actually these single-lane bikes that you can't drive a truck through them. So for donated trucks, they can get you up to a certain point. And then after that, the distribution happens by motorcycle because cars simply can't go through. So what we did, we did a couple of things. One is those roads typically are temporary because during monsoons, during rains, those roads actually disappear. So you need to dynamically have mapping, which is something that the normal— even if you go to Google Maps or if you go to OpenStreetMaps, you can't have that sort of access to a dynamic map. We also know how the motorcycles are being used. We understand better how the riders use them, how much fuel they use. We can do preemptive maintenance. So this way, by doing all this work, we can get medicine to the people who need it a lot easier and a lot quicker, and we can more or less guarantee that delivery. Along the same lines of doing open source, we developed OpenXC, which basically allows you to check the status of about 400 signals in the vehicle through a simple API. So normally, if you— the vehicle has this OBD2 port, so it's onboard diagnostics port, which is mandated by the government for emission testing. And typically you get about 20 signals out that are mandated by the government. Ford is putting out about 400 signals out there. The problem with that is that you need to understand how can the controller area network in the vehicle works. So you need to understand what every bit means, how how often the signal comes. It's basically something that really just someone who either has a lot of time to reverse engineer it can do it, or an engineer that works for that company. And it varies from company to company. I mean, we do it differently than Toyota, differently than Honda, different than anybody else. So what we did is we did all that, took all that translation upon ourselves. So now if you want to know the speed of the vehicle, you simply send a call to OpenXC from your iPhone through an app, and then you get back the reading for your speed. You don't need to know anything about the protocol. You don't need to know anything about which bit indicates what signal. You simply issue an API command and you get this information. So by doing this, we basically have allowed not only enthusiasts to play with their car and find out And we ran a whole bunch of hackathons where people did all kinds of interesting things to the cars. But also internally, it has allowed us to save a lot of time. So even internally for development, we've done a lot of work where if you need to know some signals really quick, you just use OpenXC and you get the data out immediately just by using an app. And of course, the big one is autonomous vehicles. Again, I think what's important to mention here is the definition of an autonomous vehicle because there's a broad range of definitions. But we want to go by the SAE definition of levels. So at one extreme you have level 0, which is basically a car that has no electronics in it. Level 5, it's basically full autonomy. You don't need a steering wheel. You don't need gas pedal and brakes. The car simply takes you from point A to point B, from start to destination. Just to get a sense, Tesla is like level 2.5 basically, somewhere in there. So we want to reach a level of, you know, level 5 as we announced it by 2021. And we've been really working on this for the last 10 years or so starting with DARPA projects. So we have a long history in developing autonomous vehicles. We just want to make sure that when we release them to the public, they are safe, they are vetted, that we have the right amount of sensors in them but not too much so that they are affordable for everybody. So the whole idea here is to release a vehicle that's affordable and safe so that people can feel comfortable in owning it or getting into a car-sharing situation where the vehicle is a Ford.

So where is the autonomous driving, I mean, where you do the research?

Is it in Dearborn or is it in Cupertino? It's throughout the company. So in Palo Alto typically we have slices of various technologies that we work on. That's why we're integrated with the rest of the Ford. So from Palo Alto we do mostly Sensors, and mostly next-generation sensors. So we're looking at next-generation LiDAR, next-generation radar, cameras, sensor fusion. So getting all these sensors integrated for the next generation. For the current generation, that's why you saw the announcement with Velodyne. We're still gonna use existing technology for LiDAR just because that's the only reliable technology that we have right now. We also believe strongly that you need to have LiDAR in order to So for the time being, that's our suite of sensors that we think you need— LiDAR, radar, and cameras— in order to have a complete and safe vehicle.

So a lot of vendors, they try to use the solution to implement level 2, 3, 4, 5. But why Ford is so confident to directly go to 5?

Because we think level 3, and to some extent even level 4, is not necessarily safe. And that's because you would expect a human to take over after a long, very long period of inattention. So the idea that the car drives you for, let's say, 45 minutes, and all of a sudden you have to take over for the last 5 minutes or so, it just doesn't make good sense. A lot of sense to us is that you're not going to pay attention. And we've done studies to back this up. You're not going to be in the state of mind where you can do this. You may be asleep. You may be preoccupied with something else. So that transition going from full autonomy to manual control, we don't think it can be done in a safe manner in all situations. So that's why we want to skip from level 2 where we just help the driver Directly to level 5 in order to be able to make sure that we're in full control of what goes on.

So the car has already gone on the road, started road testing?

We are testing it in Dearborn, and we've been testing it for a while. We just got, at the beginning of last year, we got our license to test in California, and we'll be testing here soon. So we're looking at different urban environments because it varies. I mean, we looked at Different urban areas, and each one of them has a completely different traffic pattern. So you need to adapt to that in the individual driving situation.