Simanaitis Says
On cars, old, new and future; science & technology; vintage airplanes, computer flight simulation of them; Sherlockiana; our English language; travel; and other stuff
AN AUTONOMOUS CAR FUTURE
SCIENTISTS AT MIT, Stanford and the Singapore-MIT Alliance for Research and Technology have explored implications of the driverless, autonomous car. Their paper, “Toward a Systematic Approach to the Design and Evaluation of Automated Mobility-on-Demand Systems: A Case Study in Singapore,” is rather more than sci-fi. The researchers, headed by MIT’s Kevin Spieser, devised a theoretic model of autonomous car use and then applied this to the city-state of Singapore.
A Google self-driving Toyota Prius. Image from www.selfdrivingeconomy.com.
In its utmost development—pushing today’s state of the art—the autonomous car is capable of completely controlling itself on its journeys from place to place. This, the researchers conjecture, would offer an enabling technology for widespread car sharing.
We’ve already seen the Zipcar idea of car sharing, where computer-arranged rentals are picked up and dropped off at other than traditional rental locations. What Spieser and his colleagues envision is more ambitious: Pools of shared cars would be located strategically around population areas, each vehicle capable of completely driverless operation.
The user/driver/sharer would summon a car by cell phone to a location of choice, use the car until it’s no longer needed, then release it for another rental or return to its pool.
There would be no need for the user to visit a rental office. Researchers aim for a scant 3 minutes from cell phone booking to vehicle driveaway. There’d be no prime real estate dedicated to extended parking at any destination. For many people living in heavily urbanized areas, it would all but eliminate the need for a private car (which typically remains inactive for more than 90 percent of the time).
Spieser and his colleagues call this Automated Mobility-on-Demand, their paper offering guidelines for sizing such an AMoD fleet based on real transportation data. Specifically, they applied their theoretical model to Singapore, an island city-state of 5,400,000 people inhabiting an area of 276 sq. miles (about one-quarter the area of Rhode Island).
Singapore skyline. Image by Merlion444.
Singapore shared data from its Household Interview Travel Survey as well as taxi records. Based on these, researchers partitioned the city-state’s road network into 100 regions, each corresponding to an average driving time from booking to driveaway of no more than 2.3 minutes (leaving 0.7 minutes for the cell phone booking). They then identified optimal fleet size taking into consideration the time-of-day effect on demand. To rectify where cars are, versus where they might be needed, the AMoD system would perform a rebalancing operation repositioning cars if needed every 30 minutes (one of the benefits of driverless autonomy).
Average wait time, for systems of different sizes, over the course of a day. Image from K. Spieser et al.
A fleet of 200,000 autonomous vehicles distributed among Singapore’s 100 mini-pools could meet a 3-minute goal—but only during non-peak demand. During morning and evening rush, the wait time would soar to an unacceptable hour or more. Increasing the fleet size to 250,000 would reduce the wait to around 30 minutes (matching typical congestion delay during conventional rush hour). A fleet size of 300,000 would reduce the maximum wait time to 15 minutes.
To put this AMoD fleet in perspective, there are now around 780,000 vehicles operating in Singapore. That is, a fully functioning AMoD system could cut Singapore traffic to 38 percent of its current state, with a concomitant reduction of congestion.
The researchers also performed financial analyses, including explicit as well as hidden costs of mobility.
See http://goo.gl/kgG1ea for access to the paper (which, by the way, is an excellent example of mathematical modeling).
Also, my thanks to journalist Ivan Berger for alerting me to a summary article by Rebecca J. Rosen in The Atlantic, “If Cars Really Could Drive Themselves, How Many Would We Need?” See http://goo.gl/tVUoaY.
If all this strikes you as cold and unemotional, consider the following: Getting the appliance drivers of the world into AMoD would leave the roads less congested for those of us with lingering enthusiasm for the automobile. ds
© Dennis Simanaitis, SimanaitisSays.com, 2014
Simanaitis Says 
Dennis –
Many cities have, in addition to ZipCar, Car2Go. Car2Go uses Smart cars. The cars are accessed with a card representing your account, linked to a pre-arranged credit card. Unique with Car2Go is that parking is prepaid and universal. A smartphone locates the cars for you, and you can simply walk to the particular car or even reserve it ahead. We use the service often in Seattle. No, they aren’t autonomous, but they represent some very fresh thinking.
An issue that I see with car sharing is that with my own vehicle I can leave an assortment of items in the trunk, even overnight. With a shared car I would have to lug these items with me each time the shared car is returned to the system.
Another issue, that unless the vehicle is returned to a depot for cleaning after each customer, there is a chance that you would get a less than satisfactory “ride”, due to other customers not caring about what they returned.
My two comments above are similar to my experiences with public transport. I do use it sometimes, but find that seats with torn upholstery and boot marks where other inconsiderate people have rested the feet, and other messiness makes the concept rather unappealing as a substitute for my own personal vehicle, where I have some control over the internal environment.