While Google may be hogging internet traffic with news of its self-driving cars, the reality is that they are forbiddingly expensive. The system, which is mounted onto a conventional car and uses cutting-edge 3D laser sensors, costs between $75,000 to $85,000 each – more than the car itself.
Now, a team in Singapore has successfully developed their own self-driving car prototype at under half the cost – just S$30,000 ($23,500) for two off-the-shelf Lidar sensors and an onboard computer which are mounted onto a conventional vehicle. That’s like a Toyota to Google’s Mercedes Benz, figuratively speaking.
The team, comprising of 12 researchers from the National University of Singapore (NUS) and the Singapore-MIT Alliance for Research and Technology (SMART), hopes to bring the cost down to S$10,000 ($7,800) eventually.
“What we’re trying to do is to forget the top-of-the-line sensors, and figure out what is the minimum we can get while still guaranteeing that the car is able to operate safely and reliably,” said Emilio Frazzoli, lead investigator of the future urban mobility group at SMART.
In a situation where all legal obstacles are removed, the team believes it can bring their prototype onto public roads in two years.
This is not the first time NUS and SMART have rolled out an autonomous vehicle. In 2011, they retrofitted a golf cart and tested it on NUS grounds.
According to Frazzoli, students initially reacted with amazement whenever the cart roamed around and emitted annoyed R2D2 sounds when someone blocked its way.
But now the vehicle has become such a common sight that people hardly give it a second glance. It’s a sign that consumers could soon treat driverless cars as an ambient part of their environment.
A potential multi-trillion dollar industry
Autonomous vehicles are being promoted as the next great advancement in automotive technology, with tremendous benefits for consumers. It frees people from the drudgery of driving in heavy traffic, which means more time for work and play.
Artificially intelligent drivers are also safer than human ones, since they won’t fall asleep, aren’t ruled by emotions, and have a wider field of vision.
Car owners could monetize their vehicle when they’re not using it – which is about 90 percent of the time – by offering it up to a car sharing service.
The vehicle, in theory, could drive the owner to work, then proceed to drop her kids at school. As part of a car sharing network, it could then pick up other passengers, even deliver goods on behalf of an e-commerce company, and then circle back to pick up the owner just as she leaves the office for home. The car owner would then be paid a fee for offering up her car, with an insurance policy borne by the car sharing company to protect against damages.
It fixes the ‘rebalancing’ problem, in which supply of shared vehicles are limited since it would be difficult to get the car to the next customer once the previous user drops off the vehicle.
Self-driving cars have implications for urban planning too. As these vehicles are utilized more, the need for parking space will be reduced.
Some forms of public transport, such as the Light Rail Transit (LRT), might become obsolete. As LRTs are fixed and inflexible, certain stations may become under-utilized. Having a network of self-driving vehicles would solve this problem as the routes they take are dynamic.
In a presentation at the EmTech Singapore conference, organized by MIT’s Technology Review magazine, Frazzoli explained the significance of a study he conducted with a few colleagues that quantified the benefits of an autonomous car-sharing service.
Using a series of simulations that combined route optimization algorithms with Singapore’s vehicular movement data, the researchers found it’s possible to cut down the country’s car population – consisting of both private and public transport – by over half from 800,000 to 300,000. That is if Singapore residents rely purely on autonomous car sharing services with an average waiting time of less than 15 minutes during rush hour.
Not yet reality
NUS professor Marcelo Ang, a collaborator on the project, says it’s possible in the future to turn the prototype into a viable car sharing business. While pricing models will still need to be worked out, it could potentially offer their services at half the price of a typical taxi ride in Singapore.
But that’s a long way from reality. While the prototype could be technically capable of moving safely on the open road, there are hurdles.
First, it’s not legal for a driverless car to operate outside of enclosed areas. Second, the team isn’t yet confident that the car is ready for an open environment, and so they’re seeking more controlled areas to further test their system.
Island resort Sentosa, university campuses, or even hospital grounds are possible places. The team is still in the midst of discussions as to where their next testing ground might be.
The Singapore government has expressed interest in the project, and officials have been speaking to the team and tracking their progress.
The state has several implications to consider. As robots take over the steering wheel, there will be side effects. Taxi drivers, for example, would be put out of jobs, which means the government has to find ways to re-skill these workers to keep them employed.
And aside from public transport policy, there are legal considerations too: how should authorities decide if a person qualifies for operating an automated vehicle? Who exactly is responsible whenever an accident occurs involving a driverless vehicle?
Nonetheless, the hurdle is not as high as people think. Legally, it may still be possible to have a driverless vehicle on the roads as long as someone is in the driver’s seat.
Cars today are already computerized to a large extent, with cruise control, electronic stability control, and anti-lock braking systems automating the driving experience.
“We see autonomous vehicles as the ultimate form of a driver assistance system, where you drive it by pressing a button and it goes…this might enable more people to use cars, such as old folks and the disabled,” says Ang.
Making driverless car sharing viable
Autonomous car sharing services will need to figure out many aspects of the business. Driverless cars, which could run purely on electricity, will need a unique way of connecting itself to a charging station – one possible way is to do it via magnets.
The vehicles will also need to be secure from malicious hackers, which means redundancies must be built into the car in the event of a malfunction. Payment systems will need to be figured out, although existing car sharing schemes may have insights into that.
Costs of self-driving vehicles will need to be reduced, and ultimately that involves moving away from retrofitting cars with equipment – a pain point for Ang and team – to having automated systems built-in by the car manufacturers.
But according to SMART urban mobility team member Daniel Morten, who also spoke at the EmTech conference, car makers are simply not incentivized to invest in driverless systems because of the classic innovator’s dilemma, where groundbreaking technologies could cannibalize a behemoth’s existing businesses.
And besides, cars of the future will become less about hardware and more about software, which means car makers would lack the expertise to successfully reimagine vehicles even if they want to.
Google, on the other hand, has been building apps for a decade, and it has the motivation to expand into the automotive business, since its revenue model is about getting as many eyeballs as possible onto its ads. This means the less time consumers spent driving, the better.
So if given a choice, Morten would rather bet on software companies like Google and Apple to disrupt the car industry.
“Also, from a financial perspective, software companies will succeed. Even the market cap of the largest car companies pale in comparison to the size of software firms like Apple,” he said.
(Editing by Paul Bischoff, lead image: SMART)