Those of us who have been following the subject of autonomous driving for years knew that the world was not yet ready for completely autonomous vehicles in wide-open road conditions. It’s too bad that a Tesla driver had to find that out the hard way recently in Florida. The car’s AutoPilot program did not recognize the shape of a flatbed trailer, apparently, and therefore did not apply the brakes. The non-driver from Ohio was killed.
There are some situations in which fully autonomous vehicles can be useful but they are limited. Big semi-tractor trailers driving from Omaha to Denver, say, could be put on auto-drive from the edge of one city to the edge of the other city. The interstate highways are marked well-enough to allow that. The Google vision of autonomous vehicles that are self-driving may work in controlled environments such as a senior citizens community or a medical complex or a shopping mall. But there is no way Google’s vehicles can survive in wide-open road conditions.
I first witnessed the problem a couple of years ago in downtown Detroit in a Toyota Lexus that was being self-driven for part of a loop around the downtown, but a driver had to be present. The reason was that the highways were not always well-marked and there were fundamentally confusing situations such as a freeway exits. The sheer complexity of what happens in an automobile as it engages with the world around it is too great for any software program to manage completely, at least for the foreseeable future.
The key question is, as the car takes on more functions, what does the driver do with his or her time? And if the car can detect a situation it cannot manage happening in the next five seconds, how does it alert the driver in a way that allows the driver to make crucial decisions? If a driver is lost in a movie and the car suddenly starts flashing red lights and ringing chimes, how fast can the human driver establish control of the situation and make decisions? That’s a problem no one has been able to solve yet.
So I was reflecting on all this as I drove a 2016 Cadillac CT6 down Interstate 87 from upstate New York this past 4th of July weekend. This car is packed with the very latest technologies that General Motors has to offer. (The company lent the vehicle to me.)
Here is a view of the dash and front seat (the driver’s wheel is upside down):
It’s fascinating to see how all the advanced features have been built into this highly luxurious package. The rear vision mirror can be switched into an actual rear-vision video camera, but I found that slightly confusing. But I loved this display (below) because the company has been able to locate sensors that capture a complete 360 degree image of what is happening on every side of the car. And the rear vision camera here is super sharp and clear.
Now, where we get more directly into semi-autonomous features is that the car is equipped with lane control, if you turn it on. Driving 75 mph, I tested it by slowly drifting into the next lane without turning on my lane change signal. The car very gently guided me back into my lane. It took over.
But if I turned more aggressively and decisively into the next lane, without using my turn signal, it allowed me to do so. It calibrated its behavior on the basis of mine. Here’s what the dashboard looks like up close. It probably was not smart to take a picture of my dashboard at 75 mph, but I felt comfortable and confident in doing so. I felt protected.
The little green icon with lane markings is what tells me I have the lane drift system on. And then the other little green icon is adaptive cruise control. As long as these two things are on, I am going to go in a straight line and I’m not going to hit anything.
The adaptive cruise control will, I am assured, bring the car to a complete stop if the sensors know it is about it hit something. I elected not to put the car to the test because I valued my serentiy too much. But I did witness the early stages of what happens if you come too close to something. I was on a local road and the car in front of me was turning right. I could anticipate that the guy was going to clear the corner, so I didn’t need to slow down.
Well, the car didn’t know that. All of a sudden, something in my seat beneath my hips started growling at me and making moves that I could feel. Big red flashing lights came on the dashboard. The system was in the early stages of a sequence that would have led it to apply the brakes. The situation passed and I did not experience that. It might not have been pleasant.
Long story short, this is as close to autonomous driving as I have ever been as a driver. On the basis of today’s technologies, there is no way the driver can be completely disconnected from decision-making. These technologies should make vehicles safer and that is a noble goal in and of itself. But Tesla has gotten ahead of itself. If it does not heed the lessons that major auto manufacturers are learning, it could be in for many other deadly incidents.
