I last addressed this subject back in 2011. Here's the link: http://sigma5.blogspot.com/2011/03/robot-cars.html. In reviewing what I wrote then I must say I did a very good job. I first covered potential alternatives to the car (the alternatives work less well than many think they will). Then I addressed engines and fuels (I think gasoline powered hybrids will be the best choice for a long time to come). If you want to know more about either subject I refer you to my earlier post. The last subject I covered then was the one mentioned in the title of both that post and this one. I made some predictions about robot cars then that are looking pretty good now. In this post I am going expand on what I said about robot cars then and add some new predictions about how things will develop.
Not a lot of time has passed since my previous post. But cars are now evolving at warp speed compared to the historical norm. The first car I drove was a 1950 Plymouth. It was one of the first cars to sport a new and, at the time "cutting edge", feature. Most cars built even a year earlier required you to stomp on a button on the floor to get the engine to turn over and start. This car contained a technological advance. You could make the engine turn over and start by giving the ignition key an extra bit of twist. The car my dad bought was the cheapest Plymouth on the lot. That car did have the new feature but it did not sport what were then options including electric turn signals, a clock, or a radio. But many other cars of that time did come equipped with all three. So, if you had bought one of those other models it would have driven just like any "stick shift" car made in the next 50 years.
While I was in college in the late '60s I drove a '58 Pontiac. It had an automatic transmission (and turn signals, a clock and a radio). It turns out that driving that car was pretty similar to driving any "automatic" produced in the next 50 years too. But in roughly the last decade cars have started changing quickly. A couple of years ago I drove a car that literally had a push button. You push the button to tell the computer that you want to start driving (or that you are done). That's a really different experience than the old "put the key in the ignition and go through the rest of the rigmarole necessary to get the engine to start" routine. In fact with many cars now you don't even have to put the key in the ignition at all. As long as it's in your pocket you are good to go.
And that's not even the current cutting edge. A friend of mine has a Toyota Prius. It has the same push button (and key in the pocket) setup but what happens when you push the button is now radically different. Now he no longer controls what is happening with the engine. The computer does. He pushes down on the "gas pedal" but now all that does is tell the computer you want to go faster. The computer figures out how to make it so. What's going on?
I recently saw a story about a $9 computer. The basic package does not include everything. There's no power supply or keyboard or monitor. But the same people that sell the $9 computer will sell you an add on that includes all this stuff and it costs $15. So you can get the whole package for about $25. Why does that matter? Because that $9 device includes a CPU, RAM, a solid state disk, and a number of standard ports you can plug stuff into. It is powerful enough to run real computer programs. And it costs $9 if you buy them one at a time. My point is that large amounts of computer power is now really cheap.
That means that car companies can afford to put a powerful computer in your car. And that computer has access to power and it will be designed to get inputs from things like the gas pedal or various sensors scattered round the car and it will be designed to send outputs to things like controls on the engine or actuators on the transmission or whatever. So to add a computer to a car costs the car maker $9 or possibly quite a bit less. And, if you can use the computer to replace something like the old complex machinery that figured out when to shift automatic transmissions you can save the cost of the components you are replacing. In fact, replacing components with a computer can actually save a lot of money. The computer has a negative "cost".
The tipping point on all this was reached about 10 years ago. I currently drive a '99. It has a limited amount of computer capability. But the same make and model of car made only five years later has several much more powerful computers, each dedicated to a different task. That is now just the cheapest way to go for the car companies. And computers keep getting even cheaper and even more powerful.
I talked about the DARPA challenge that kicked this whole "robot car" thing into high gear in my previous piece. To pull that off then contestants had to jam many thousands of dollars of expensive computer and sensor equipment into the cars. It was hard getting everything to just fit and function. But car companies have gotten better and better at integrating computers into cars. And the sensors have gotten incredibly small, incredibly cheap, and incredibly frugal in their power consumption. Why? Because in many cases what we are talking about are smartphone components.
Smartphones are small and have tiny batteries that don't work very well. So cell phone components have to be small and have to be very frugal when it comes to power consumption. They are still relatively expensive. If you bought an iPhone 6 without most of its cost being hidden in your monthly phone bill it would cost about $600. That's a far cry from $9. But you can now get a phone with the capability of an iPhone 1, introduced in 2007, for perhaps $100. It would look pretty wimpy and pathetic sitting next to an iPhone 6. But the iPhone 6 costs now what the iPhone 1 cost then. The big difference is that the new components that are so much more powerful and so much more capable now cost what the iPhone 1 components cost then. The price of this sort of thing is plummeting at an astounding rate. And it is going to keep plummeting.
This is reflected in the features you find on new cars today. The capabilities that were available on perhaps one or two high end cars in 2011 included sophisticated cruise control, blind spot detection, collision avoidance, and collision alert when you are backing up. All of these features and more are now available on mid-priced cars. But the "same" features on these new cars work much better now than they did then. These include the ability for the car to find the lane markers and keep you in your lane and a feature that tries to figure out if you have nodded off behind the wheel. And, as the late night infomercials have it, "but wait - there's more". Car makers are rolling out new and improved features like they are in the fashion business.
One thing I pointed out in my previous post was the importance of cars talking to each other. That looked like a problem then. But the Federal Government stepped in recently and is now coordinating a project to standardize inter-car communication. That problem got solved before most people even knew there might be a problem. The rate at which cars are gaining intelligence is astonishing. And perhaps most astonishing is that several car makers have announced that they will be selling robot cars by 2020. That's only five years from now. Even more astonishing is the fact that it looks like they will hit their target date.
There is always someone willing to try to scare people. With respect to robot cars the standard line is "what if a robot car goes berserk and kills a bunch of people?" It could happen. But look at Google's experience. I was able to report in 2011 that a Google car had been in an accident. I can now report that Google cars have been in 11 accidents. And something surprising, at least to the "kill a bunch of people" crowd has happened. In all but one of those accidents it was obviously the fault of the other person. The accident that was not caused by the other driver happened while the car was being driven by a person not the automation.
The most common accident scenario has been where the Google car is stopped at a stop sign or a traffic light and someone rear ends them. The accidents were not the result of the stop. The Google car had been stopped for a while when it was hit. The other driver was just not paying attention. And all the logs and readouts were able to unambiguously and with great detail and specificity tell the tale of exactly what happened and who was at fault. The attention of robot cars does not wander. They do not text while driving. Or, if they do, they are able to multitask so the quality of their driving does not suffer. It is very likely that robot cars will make the roads safer, much safer. Robot cars also do not drive drunk. By far the most difficult problem the robot car people have had to cope with is other cars being driven by people.
I am sure that when robot cars first become available they will not be for most people. For one thing they will be expensive. And a lot of people will be unwilling to trust their lives and those of their friends and loved ones to a robot. I expect that concern to disappear pretty quickly. The cost problem will take longer. I expect the first robot cars that hit the road to be quite sophisticated. But I think they will follow the iPhone model. The first iPhone blew everybody away. Until, that is, the iPhone 2 came out. Apple has succeeded in making each generation of the iPhone more cool than the previous one.
I expect to see the same thing with robot cars. With the iPhone Apple saw how people were using them and that suggested improvements. Also, so many iPhones were sold that the parts manufacturers were able to figure out how to make the same part more cheaply or a better part for the same cost. So later iPhones had better hardware. Apple turned that into new features or old features that worked better.
I don't expect robot cars to sell in the numbers iPhones do. But advances in smart phones will help car makers out as will experience gained as many cars hit the road. I think robot car 2.0 will be much better than robot car 1.0. And then 3.0 and 4.0, etc. will all be a big improvement on the previous generation model. iPhones have been around for less than ten years. Every time a new model rolls out people ask "how can they make the new model better than the old one"? So far Apple, and the other smart phone makers, have figured out an answer that generates an incredible volume of sales. At some point this will stop. But I have no idea when it will be. I think it will take longer for robot cars to get to the point where the new models don't not seem like that much of an improvement over the old one. But I am not going to try to guess how long that will take. Instead I am going to apply my crystal ball to a other related issues.
As I covered more thoroughly in the previous article there are people that think cars are the root of all evil. I think like an engineer. Engineers don't think "cars are evil". Instead they think "what are the problems with cars?" Then, if they agree that this or that actually is a problem they go on to ask "what is the best solution?" If the best or only way to fix the problem is "get rid of cars" then "get rid of cars" becomes the engineer's solution of choice. But the focus is on solving the problem and not some moral judgment about whether cars are good or bad. We live in a world where people like to fixate on the moral judgment to the exclusion of trying to find the best solution to the problem. But instead of doing that I am going to apply the engineer's perspective to a couple of problems associated with cars. I am going to skip over the "is it a real problem" part because I think both of these are real problems.
The two issues I want to address are pollution and streets. Let me start with pollution. Cars turn gasoline into several different nasty things. One of them used to be smog. Smog is smoky, smelly, and unhealthy stuff hanging around in the air. It is unpleasant. It blocks our ability to see blue sky and fluffy clouds. It also just smells bad. So it is an aesthetic disaster. But it also turns out to be a health disaster. It does terrible things to your lungs. Smog in the US is mostly a thing of the past.
Changes to the formulation of gasoline and to how cars work got rid of the problem. Some places still have a smog problem some of the time but for most of us it is pretty much a thing of the past. The guts of a car (and the chemistry of gasoline) are now quite different than they used to be but it turned out to be possible to fix cars so that they did not create smog. China has a terrible smog problem. And some of this is caused by cars. But most of it is caused by coal fired power plants. And modern designs for cars and gasoline would take care of the car part of the problem. So China can fix its smog problem. It just has to decide to fix it and then do so.
That's an example of where the solution was not to get rid of cars but to make changes so that the problem went away. If smog was the whole problem we would be done. But it isn't. It was just the most visible and the easiest problem to fix. The big problem that everyone knows about is carbon dioxide (CO2). If you have a gas (or diesel) car it is going to make lots of CO2. There is just no getting around this. If you switch to Natural Gas as a fuel the amount of CO2 goes down but it is not eliminated. If you switch to Hydrogen you can eliminate the CO2 problem. But I indicated in 2011 what the problems with Hydrogen are. I still think they are insoluble so I don't see Hydrogen ever going anywhere.
The best you can do is to manage the problem to minimize it. There is plenty of room to make cars much more efficient. That's why I am a fan of hybrids. I think you can get the amount of CO2 per mile way down from where it currently is. A car's basic purpose is to move around. To make this work you need a way to have a large amount of stored energy move around with the car. A gas tank full of gas represents an unbelievably large amount of stored energy. It is so vastly superior in this role to anything currently available that I frankly have trouble imagining an alternative. But we can do a lot better at getting the most out of what is there. In almost all cases burning generates CO2. And lots of the burning we do does not have mobility issues. So I think there are lots of ways to drastically cut the amount of CO2 produced in these situations where mobility is not involved.
Now I would be delighted if I could justify tossing out the entire preceding paragraph out. If someone comes up with a decent battery, a device that can store and release a lot of energy at a small cost in terms of volume, weight, and money, then I would be all for all electric cars. A good battery would make all kinds of approaches to CO2 reduction possible. That would mean that we could ditch the whole hybrid thing for something much better. The good news is that between 2011 and now there has been some progress in battery technology. The bad news is that the amount of progress has been very modest. Batteries still suck. And I see nothing currently on the horizon that looks promising in terms of a large improvement in batteries. We should do what we can to reduce CO2 emissions from cars and other vehicles but we are not going to get it to zero or anywhere near zero. I wish I had better news on this front but I don't. The same is not true about the other issue.
I know a couple of people that know a lot about urban planning and that sort of thing. Both of them do not like how much of a city ends up as pavement (streets and parking). In actual use streets are very inefficient. We get in our car and drive somewhere. On both ends there has to be someplace to park the car. And we go to several places. So we don't need two places, on average, to park our car. We need more like ten. Now if you drive to the mall today but don't tomorrow the parking place you use today can be used by someone else tomorrow. But you don't go to two or ten places in your car. You go to hundreds or thousands. So, on average, many parking places have to be provided for things to work. That means vast amounts of land devoted to parking.
Then there are the streets that connect everything together. First, there has to be a street to wherever you might want to go even if you don't do it very often. Then there has to be enough capacity so that you can go there when you want. It doesn't matter if the street is wide open at three in the morning if you want to go there at three in the afternoon. So in the same way that there needs to be lots of parking spaces for each car there needs to be lots of streets to handle the traffic. So here too the average utilization is low. And that bugs the urban planners. They like alive and green not paved and barren. And most people agree with them. It seams reasonable to be unhappy about the amount of land area in a city devoted to streets and parking. It seems reasonable because it is.
The standard urban planner solution is to replace cars with feet or bicycles or mass transit. I looked at those options previously too. So now let me focus on how robot cars can make things better, possibly a lot better. I am going to start with a simple idea: If a car can drive itself it can park itself. Ok, how does this help? Consider the present situation. When you are home where's your car. It's in the garage or on the street. Why? Because you don't want to have to hike a long distance to get from your car to your home. But the need for close in parking goes away if the car can drive itself. It can be waiting close at hand when you want to go somewhere but it can be somewhere else the rest of the time. It can also drop you off then go away when you return. You don't need to have an attached garage or on street parking any more.
As long as the car is reasonably close, say within a mile, it can get to your door in almost no time. An application on your phone to summon it a minute or so before you need it is not hard to implement. Your phone, if you have a smartphone, already has much more complex applications running on it now. So that takes care of the home end. On street or on premises parking is no longer needed where you live. And the same kind of approach also works at the other end. As long as you car can be summoned quickly it does not need to be parked within a few steps of your destination. Then there's the in between part.
There are expectations about how close you will stick to the car in front of you. They are based on an average plus a margin. On average a driver can react to changing circumstances in a certain period of time. You need to be far enough away so you don't hit something while you are in the process of reacting. On top of that a margin needs to be built in because this is not the kind of thing where "mostly right" is ok. Enough margin needs to be built in so that it is extremely unlikely that two vehicles will hit each other. That means that even the most obnoxious tailgater leaves considerable space. And the timid may leave a ridiculously large amount of space.
A standard number for the capacity of a highway car lane is 2000 cars per hour. That translates to a bumper to bumper distance between cars of about 160 feet. I drive a big beast of a car. It is almost 20 feet long. If all the cars on the road are about 20 feet long that means that following each car is 7 car's worth of empty space. If we could somehow jam cars bumper to bumper the same freeway lane could handle 8 times as much traffic. With cars driven by people you can't do that. The situation on city streets is not as bad. The cars are, on average, much closer together. But the idea is the same. And then there are those pesky traffic lights. I spend a lot of time stopped waiting for traffic lights. And I can't tell you how much of that time I have spent stopped even though there is no one coming.
Remember the part where cars will soon be talking to each other. Well how about putting a "car" at each intersection. It wouldn't actually be a car. It would instead be the traffic light system. It could tell my robot car when the light was going to change. That might help. But much more is possible. Most traffic lights are connected to sensors in the road that detect cars. Pretty much the only thing this information seems to currently be used for is to decide whether or not to cycle the "left only" signal. Smart traffic lights could also use the information to retime the light cycle on the fly. This should allow the lights to cycle in a more efficient pattern.
The sensors in the road can also be used to continuously monitor the amount and timing of traffic. This would allow the signal duration to be dynamically adjusted to allow for increases or decreases in traffic no matter when or why they happen. Many lights now have a "night" pattern or a "rush hour" pattern. But these are hard wired in and only infrequently changed. It is very expensive to gather the data and make the changes so it is rarely done. And patterns are set for average or expected or typical conditions rather than actual current conditions.
So we can get the traffic lights to work better. We can have the signal let the cars know when the signal is going to change. The information can also flow in the other direction. My car can also tell the signal when it needs the "left turn" cycle activated or that it is going to be making a right turn so there doesn't need to be a "through" cycle if it's only for my car. That sort of thing effectively increases the capacity of streets. And, where it makes sense, increasing speed limits also increases capacity. If the separation between cars doesn't need to be increased then a higher speed limit increases the capacity of the street. Computers in cars do not become inattentive and they can react very quickly. So once most cars on the road are robot cars we can substantially increase (double? triple?) the effective capacity of streets. This means the same streets can handle a lot more traffic or the same amount of traffic can be handled by a far more modest street system.
And then there is this. Most people now want a car that is their own. Part of it has to do with the fact that a car is a statement. Forty year old men driving a muscle car are making a statement. But so are twenty-five year old women driving a cute pink car. The auto industry has come up with pretty much whatever kind of vehicle you need to make whatever kind of statement you want to make. Then there is the fact that some people haul a lot of stuff around. They need a car of their own because it needs to have their stuff in it. And so it goes. But then there are other people who just don't care. If they want to make a statement they want to do it some other way. Or they see a car as basic transportation. If you can get them to where they want to go they don't care how it is done. At present most of us find a car to be the most practical option. But what if that were not true? What if there were other practical options?
Some people manage to get along without a car. One trick they employ is to occasionally use a taxi. If you only need a car once in a while it is cheaper and may be more convenient to cab it. And we now have Lyft and Uber. They provide what is essentially a taxi service but it is cheaper and more convenient. At least that's their story. But we can go a lot further in this direction in a robot car environment. What if you don't own the car you summon? Say some of the cars in that garage we were talking about before are pool cars. They are available for anyone to use. If an appropriate car shows up and gets you to where you want to go does it have to be a car your own? For some people it does but for others that is not a requirement.
Robot cars allows us to take the taxi/uber/lyft model even farther. Presumably the direct cost per mile would be higher. But the purchase and maintenance costs would be zero so that actual cost would be lower. And now the number of cars it takes to make everybody happy drops substantially so the demand for parking drops and the urban planners get to reduce the amount of pavement. That makes them happy and us happy because all of us like more greenery in out lives.
And lots of people currently buy more car than they routinely need. I have a big beast. But most of the time there is only me so it is overkill. If I could summon a small car when that's all I need but get a big car for those few times that I need one then I will, on average, be in a smaller, more efficient car, than what I now use. Yet I will have something that in a practical sense works better than what I now have. This will not work for everyone. But it will work for a lot of people.
If you are an "I hate cars - period." person none of this will be persuasive. But if you are in an "I hate cars for the following specific reasons" person or if you are an "I have and use a car because there is really no practical alternative" person then I have shown you a future that looks pretty good. And it looks like robot cars will arrive sooner than I expected them to. And that's even better. I, for one,. can't wait.
No comments:
Post a Comment