The End of Cars?

I’ve been thinking about cars lately. Not because I drive much, but because I wonder what the city would be like without them.

And I don’t think that day is too far off.

Consider: the technology for driverless cars is already here. Google has unveiled a “real build” of a fully-automated car; Nissan, Audi and others have all announced they will be commercially releasing “driver assist” technology in two or three model years.  And Uber has apparently hired some 50 robotics scientists to work on automating its fleet.

Google Self-Driving Car
(Google’s adorable robo-car; photo courtesy of Google)

The gating factors for this technology are more regulatory, cultural and – most importantly – the perceived need to have cars do everything they do today, without a driver.

A thought experiment: what if a city were able get rid of private cars overnight, and replace them with a system for getting citizens around town flexibly, quickly and cost efficiently?

I know, ignore my bias against government regulation for a minute and go with me on this.

So, no more cars. You want to get around town, you need to walk, ride a bike to take public transit.

Ugh, right? Sometimes you need the flexibility and (hopefully) speed of a car.

But urban public transit could revolutionized by self-driving cars. In our hypothetical city, light rail and buses are rejected as being yesterday’s technology: too inflexible, too expensive, and dependent on an antiquated system of schedules and stops that don’t take advantage of the massive advances in communications and positioning technology that have occurred in the last decade.

Instead, the city deploys a fleet of self-driving vehicles. They come in different sizes and configurations, are dispatched by smartphone, and don’t go beyond the city’s otherwise-car-free streets. If you need to get somewhere in town (and don’t want to walk or ride your bike) you call for one of these vehicles. Think of it as a combined Uber and Car2Go – but fully automated.

Safety

Hand-wringing over robot driving aside, it’s pretty clear that any self-driving car would be massively less likely to get into an accident than the easily-distracted, sometimes-impaired wetware currently necessary to drive a car.

But our automated city system would offer safety benefits beyond even those the come along with an always-vigilant, computerized driver. Because these are the only vehicles on the road (there could be designated routes for commercial deliveries), and because they only operate within the city, they could be built very differently than today’s autos, which must be safe at highway speeds. With a fully automated traffic system, the vehicles would rarely need to stop; they could “flow” around town. This would enable relatively slow speeds; max speeds of 25 mph would still get riders most anywhere in town in 20 minutes or less due to the elimination of stopping and congestion. A 25 MPH-max car could be built lightweight and simple, able to stop on a dime. These cars would pose far less of a threat to pedestrians and cyclists (and not just because there no longer is a human behind the wheel).

Complexity

Another big objection to driverless cars stems from the assumption that such vehicles would need to operate in all of the same circumstances cars currently do. City streets, freeways, remote desert highways, all in any manner of weather conditions, and at speeds between 0 – 90 MPH. If you assume you need to solve for all THAT, you’re requiring a massive amount of complexity in the navigational, safety and decision-making capabilities of the vehicles. You’ve got to address fog, and snow, and construction reroutes, in all possible areas, and at sorts of speeds.  You’ve also got to build in safety systems capable of protecting vehicle occupants in high-speed crashes.

But limit the system to the small, tight and well-monitored ecosystem of a city’s streets (or even just most city streets), and these problems become orders of magnitude more manageable. At the far smaller scale involved, navigational markers can be programmed and updated at a very granular level; the streets would become a virtual track for the cars to operate on.  And it’s easier still, if, like our model city here, you can do so without having to deal with traditional cars at the same time.

Infrastructure

Our city would need far, far less space for parking. Most space currently used for street parking could be given over to drop-off zones, bikeways, and promenades. Some parking structures would be necessary for storage and maintenance of system vehicles, but far fewer spaces than a city with private cars would need. In fact, with the fleet of vehicles continuously in use going from call to call, our city might well have 10x fewer vehicles than a city with traditional cars.

How? Consider how little private vehicles – particularly vehicles owned by in-city residents – are used. The average commute time for Seattle residents is 25 minutes. That means that the typical driver is using their car 50 minutes a day. Even if you increase that number by 80% to account for errands, etc. you’re still left with this: our cars are sitting idle 93.75% of the time.  With smart fleet management and dispatch software smoothly sending cars from call to call, our city’s vehicles could be in use over 50% of the time (including times of low demand).

There are obviously some big environmental benefits coming from this – not just the lower cost per mile to operate simple, light vehicles at low speeds, but also the massive impact of eliminating the manufacture of so many cars.

This piece of the puzzle, believe it or not, is already being worked by Uber with “Uber Pool.” Check out Bill Gurley’s in-depth look at the math behind Uber Pool, and why it’s potentially such a huge deal.

Cost

The cost of a system such as this – whether to the city or a private operator – would be much lower than traditional public transit. There’s little infrastructure to build, and the vehicles themselves – freed of the operational complexities of traditional cars – should also be cheap, and can be scaled up as demand dictates.

For users, the cost would be far lower than owning a car. Lower-cost vehicles being utilized at 10X the frequency of traditional private vehicles compels that result. In fact, the cost could easily end up being cheaper even than current public transit options.

Take a look at some current monthly costs for transportation in Seattle:

Monthly Metro Pass – $81.00

50 15-minute Car2Go rides – $307.50

50 5-mile UberX rides – $600.00

The Uber option is already cheaper than owning a car and parking downtown.  Imagine how low that could go with a liquid, automated system utilizing much cheaper vehicles.

Access

What of the disabled, and those unable to operate or afford the mobile phone, or rides on the system itself? These are issues that today’s transit systems have to contend with, and while I can’t predict exactly what the solutions would be for a robo-car system, there would be more – and more flexible – options than exist for transit operators today. The sheer number of vehicles involved in the system offers the potential for solutions that are far cheaper and easier than the brute force approach of equipping every bus with an expensive, disruptive and prone-to-fail wheelchair ramp.

Outside the City

While this system would solve the problem of in-city transportation – essentially replacing private vehicles and traditional public transit like light rail and buses- it would take longer for driverless cars to take over outside of cities. As soon as you try to do that, you’ve got to solve for higher speeds, lower density and more complex navigation.

But let’s call that a second-order problem. To get started, our urban robo-car system could simply interlink with the outside ecosystem of traditional cars. Those coming in from out of town would park in lots – like today’s park-and-ride lots – where they would grab a driverless vehicle for the trip to their in-city destination.

Obviously, a system like this would require sophisticated algorithms for traffic balancing and flow. Minimizing wait time and cost would be critical to adoption. But the technology and math know-how to do this? We’ve already got it. Putting it in place is really just a matter of overcoming a century’s worth of how we think about cars – and the jarring shift to get there.

Leave a Reply