Raj Rajkumar: Driverless Vehicles

Raj Rajkumar: Driverless Vehicles


I got into this area of self-driving vehicles
in 2006. I was running a lab for General Motors, looking
at embedded computing systems in a car. It turns out that the concept of a driverless
vehicle, a vehicle that drives itself, dates back to the 1940s when General Motors was
the biggest car maker in the world. And during a World Expo, they actually proposed
this idea where a family in a car could be sitting around a table inside the car, talking
to each other, playing, maybe playing card games, and the vehicle is driving itself on
the highway. So the idea of a self-driving vehicle is many,
many decades old. It turns out that we at Carnegie Mellon have
been working on driverless vehicles since the mid-1980s. We were the very first ones to prototype early
generations of driverless vehicles. This car that we have built is the 14th generation
of a self-driving vehicle built at Carnegie Mellon University. Also, we have a 2011 Cadillac SRX, a crossover
SUV, that is being fully automated, it can drive on highways, at highway speeds, even
70 miles per hour. It can take an exit ramp, it can actually
take an entry ramp, it can merge, it can change lanes. It’s capable of doing all of that. It can actually drive in urban settings, suburban
settings, change lanes in a suburban environment for example. It can actually know the status of traffic
lights, stop when it needs to. It can actually go when the traffic light
turns green. It understands work zones. It can recognize pedestrians, bicyclists,
and then stay away from bicyclists and pedestrians. Or follow them at a slower speed if need be,
and stay away from cones and barrels. So it’s already capable of doing a lot of very impressive
capabilities. There are multiple reasons why driverless
vehicles will make a big impact on society, will revolutionize transportation as we know
it. First and foremost, about 1.2 million people
die every year due to automotive accidents globally. All these fatalities, accidents, and injuries
add up to an economic cost of more than half a trillion dollars, U.S. dollars, every year. Most of these accidents are due to human error,
human distractions. Humans can be distracted very easily. So if we can take humans, if you will, out
of the driving equation, we expect that these fatalities, injuries, and accidents will drop
dramatically over time. So really, our holy grail, if you will, is
to basically take the number of fatalities from automotive accidents towards zero over
time. Number two, the average American, today, commutes
for about 51 minutes to and from work. And about 70% of cars have a single driver
in them going back and forth, so it’s basically 51 minutes of unused time, unproductive time
just getting from point A to point B. Thirdly, there are elderly people and disabled
people, who because they have lost their cognitive capabilities due to old age, or because they
are legally blind, or physically blind they cannot drive. They have lost their mobility, therefore they
lose their independence, and therefore their quality of life has shrunk dramatically. This could actually basically make a huge
change in lifestyle for those people. And the fourth factor is that pollution and
environmental effects from automobiles are significant. If cars were driving themselves, they likely
coordinate, communicate with other vehicles on the road, all of them would be driving
at kind of the same speeds, what is called harmonized speeds, and therefore everybody
makes forward progress at about say 50-55 miles per hour uniformly and our commute times
would get shorter, and therefore the environment will also benefit. There are two big practical questions that
people always ask. One is, how much is it going to cost? Secondly, how is this going to look? So, let’s talk about the cost issue. For this technology to be adopted it has to
be affordable; it has to be maintainable. And we expect that the technology for doing
this later this decade, early 2020s, is evolving, maturing very rapidly. We expect this capability will cost an additional
$3,000 to $5,000. Let’s talk about the looks of the car. They cannot look like engineering prototypes
and you need to be comfortable with saying that this is my car, that have the right color,
the shape, the looks and such. So we expect that self-driving cars of the
future will look normal on the outside and the inside. So when we built our 14th generation autonomous
vehicle at Carnegie Mellon, our self-driving 2011 Cadillac SRX, we made sure that it does
look aesthetically normal even today on the outside and the inside. Over time more and more of these capabilites
will fall into place, and one fine morning, the vehicle will be able to drive itself completely. But that last step, will be a relatively small
step, will be very natural, very organic, very incremental. And even though it looks like an evolution
today, it would have been a very simple process of evolution, and bingo, at that point in
time, we are being driven around by driverless vehicles.

One thought to “Raj Rajkumar: Driverless Vehicles”

Leave a Reply

Your email address will not be published. Required fields are marked *