Thursday, October 20, 2011

Autonomous self-driven vehicles

Stanford's autonomous car. (source)
About fifteen years ago, it seemed ridiculous that everyone would be running around with a personal communication device in their pocket that would allow access to anyone else in the world as well as the Internet. Right up there with flying cars. The rapid increase in computer technology paired with a decline in prices allowed it to happen.

In a similar fashion, I think we are seeing the beginnings of true autonomous vehicles. That article is a little old now, as the problem has been attacked for a while, and even has had some high-profile demonstration solutions from Stanford, CMU and Google. I remember seeing the Red Team's Hummer sitting around when I was working nearby, back in 2005. It seems that people have been quietly working on other projects since then. It's one of those problem that can be effectively addressed by throwing more computing power at it -- which means it will get easier over time. And it could result in such a huge convenience that -- like cell phones -- we may wonder how we lived without it.

Leaving the technical details of implementation aside, what are the implications for public transit if everyone has access to an autonomous vehicle? Keep in mind, I think the first and most obvious application for this is cheap driver-less taxi. It will be as if there was a private bus that operates on exactly the route you need, anytime you call for it. Will that eliminate demand for public transit? It certainly sounds convenient. There are a couple problems.

Will the public be comfortable with autonomous vehicles? Probably not at first, even absent legal issues. It is likely that the computer controlled vehicle will be safer than having a human driver, but there will be an instinctive resistance to the idea. There could also be some confusion for pedestrians as they cannot look inside the car and read the face of the driver -- maybe indicator lights on the front could resolve this. However, there are several big potential benefits for pedestrians. Eliminating distracted or impaired drivers is one. More subtly, the need for proximate parking is reduced by sending cars to remote lots, or even obviated with extensive car-sharing. This is one path to lifting the scourge of automobile storage off the streets of cities.

Introducing many vehicles onto the road will cause a congestion problem. The autonomous vehicles can probably be smaller, and pack more tightly using by guide-ways, but it will reach a limit at some point. Larger vehicles such as buses will still be the kings of capacity. Of course, without having to pay and accommodate a driver, buses can run many more and varied routes, at all times of day. Frequency will be limited not by operational costs, which will drop significantly, but just by capital costs of acquiring a sufficient fleet. It could also be the case that software may identify outstanding transit requests which largely overlap, and then dispatch a van or a bus to handle multiple passengers. If such a service costs less than ordering a private vehicle, I can see many people signing up for these "car-pooling" systems. It could even suggest that a permanent bus route is appropriate if the same demand is issued along the same route every day.

Then there is the matter of fuel efficiency. If we continue to use the internal combustion engine, both the direct and indirect costs of operation of these vehicles could still prove to be prohibitive. Luckily, autonomous vehicles are well-suited to electric operation. They can be programmed to find themselves a parking spot with a power plug when not in use. It is even possible to conceive of electrified roadways that function similarly to electrified railways. With computer guidance, it should be possible to safely transition from being an automobile to a "trolley" and back. These kinds of roads could even make long-distance all-electric autonomous operation possible.

Of course, by that time we could also have high-speed railways linking cities with much faster speeds than any road vehicle could safely achieve, and airplanes for further distances. The last mile problem would already be solved by the presence of driver-less taxis and other transit vehicles. Unless you are particularly attached to your personal car, there is no reason to spend all that extra time riding in it, when you can have all the same convenience wherever you go without it.

Of course, there will always be people who want to drive their own cars. Fair enough, I used to be one of them, I know how it feels. I am not sure how the transition will work, but I can't imagine that people will be too comfortable with autonomous vehicles at first. There could be restricted autonomous-only roadways. Or perhaps the technology is so successful that it can safely drive and accommodate being on the road with humans. In the latter case, there will probably be intense pressure to replace the legacy vehicles with new ones capable of autonomous operation (in addition to manual operation, perhaps). I imagine there will be a rise in private motorways where people can manually drive cars either for sport or enjoyment, without fear of traffic congestion, while leaving the day-to-day grunt work of dealing with public roads to computers.

I know this whole article sounds like pie-in-the-sky talk, but I think that autonomous vehicles, more than anything, will transform our society and cities. One worry is that we may end up repeating the mistakes of the 20th century and try to develop Hypertrophic cities all over again, like something out of Metropolis. But hopefully we will have learned better this time around, and instead try to take advantage of the characteristics of self-driving cars in ways that can alleviate their impact on the city.

No comments:

Post a Comment