Imagine all of Singapore’s mobility needs being fulfilled by a fleet of 300,000 autonomous cars. That includes all journeys made by cars, taxis, motorcycles, buses and trains, amounting to 12.5 million trips per day (2012 figure), being catered to by one-third of the total vehicle population here.
This hypothesis was presented in a scientific paper written by academics at the Massachusetts Institute of Technology. Kevin Spieser, Kyle Treleaven, Rick Zhang, Emilio Frazzoli, Daniel Morton and Marco Pavone painted a scenario in which 200,000 to 300,000 autonomous vehicles will be sufficient to replace some 900,000 road vehicles and a railway network spanning nearly 180km.
Users will book a vehicle – much like people book cabs today, but probably via a smart device – and the nearest available one for the designated journey will arrive. The paper supposes that a fleet of highly autonomous vehicles can be as small as just below 93,000 for this to work. But then, waiting time will be inordinately long, especially during peak hours. A fleet size of 200,000, however, will lead to more acceptable waiting times. And 300,000 vehicles will bring average peak waiting time to less than 15 minutes.
The paper states that the average private car today is used less than 10 percent of the time each day. It adds that car-sharing schemes have limitations, such as availability of docking points. In contrast, an autonomous vehicle in a properly sized fleet will probably be utilised round the clock. And there is no need for a docking point because any available parking space will do.
The authors feel that Singapore is ideal for such an experiment because of its land space constraints, limited potential for road expansion, and its dense development. They add that Singapore has access to a rich data mine, via EZ-Link card transactions and taxi tracking, making it easier to plan for an autonomous fleet.
If the paper’s proposals can be realised, the impact will be game-changing. It means little or no traffic congestion on the road, and no more multi-billion-dollar plans to expand the rail network. But what about cost?
The writers estimate that converting each car to be autonomous can be done at a cost of S$20,000. The lifespan of each car, however, will only be two and a half years – presumably because of the extraordinary high usage, even higher than that of taxis today.
But this will be offset by savings elsewhere. For instance, the paper asserts that an autonomous fleet affords high value of travel time savings. Presumably, this is largely because the traveller can do other things besides driving during the entire journey.
“Private vehicle owners in Singapore spend, on average, 458 hours per year driving in their car,” the authors wrote. “Factoring in the time for parking and other related activities brings this total to 747 hours per year.”
After factoring in this time, which can be monetised, the total mobility cost of a shared autonomous vehicle is roughly half that of a conventional human-operated car. The paper says the savings represent about one-third of GDP per capita, which is quite significant.
The paper presents an extremely attractive proposition from an economic point of view. Autonomous vehicles are many times more efficient than private passenger cars. But in reality, how feasible is such a fleet?
Well, technologically, such vehicles have already passed the proof-of-concept stage. Semi-autonomous systems are already found in cars today. These include self-parking, adaptive cruise control and steering assist. From as early as 2020, companies such as Nissan, Volvo, Daimler and Google will have fully autonomous vehicles ready for market introduction.
Technical feasibility is just one aspect of going autonomous, though. There will be legal issues and matters of insurance coverage. For instance, if a driverless car gets into an accident (unlikely, but still possible), how should insurance claims be settled? If one gets into a fatal accident, who is culpable – the occupant or the manufacturer?
There is also the major issue of ownership. In the MIT paper, the fleet is shared. So who owns the vehicles? In theory, there could be one central agency that owns and maintains the fleet, and consumers pay according to usage. But which local agency or body is willing to undertake such an enormous task?
The two biggest challenges, however, have to do with human nature. One, do we want to share a car with strangers? Past attempts here to get commuters to share cabs fell flat – because people view taxis as a close proxy to the private car. So, they do not like sharing. Two, there are people who actually love cars and driving. Having autonomous cars will not go down well with them. Perhaps we can still allow some privately owned, personally operated vehicles, but set at very high prices so that only a small percentage of the population will be able to afford them. Otherwise, having too many of these vehicles may undermine the efficiency of driverless cars.
So, while the MIT paper illustrates the exciting possibilities of autonomous cars from an urban transport perspective, the reality is often more complex. In probably more ways than one, you could say that the biggest barrier to driverless cars in Singapore are, well, drivers.