Two and a half years ago a small group of graduates from the Massachusetts Institute of Technology (MIT) decided to take on the biggest issue the autonomous car industry faces: urban driving.
Naturally, testing an autonomous car on the busy streets of Boston, or even in Silicon Valley, was going to be tough. Especially if they wanted to work in daylight, so the team started looking abroad before landing in the south-east Asian city of Singapore.
Chief operating officer Doug Parker tells Techworld: “Our founders Emilio [Frazzoli] and Karl [Iagnemma] flew here to set up our research and development team. That’s really why we’re here: it’s the talent.”
“We have been amazed by the talent we have seen over here. At the moment, we are the only people in town doing this, in a few years it’s going to be different. Right now we feel like Singapore is our secret place. We’ve got a great team, the government supports it, the roads are good.”
Just this week Singapore's foreign minister Vivian Balakrishnan spoke about how he wants the city-state to be a hub for autonomous vehicle startups.
How about cost? “Cost isn’t really an issue,” says Parker. “It’s an R&D project so we will do whatever it costs to get the best car on the road.” NuTonomy has raised US$3.6 million in seed funding so far, led by Fontinalis Partners, Samsung Ventures and Signal Ventures.
How does it work?
NuTonomy is concentrating on unique software and logic and applying it to off-the-shelf hardware right now, rather than an all-in-one solution. NuTonomy has applied its sensors, cameras and software to a Mitsubishi i-MiEV electric car for the prototype, pictured above.
The software works on algorithms which direct vehicles around complex urban roads through precise GPS mapping of streets and localisation using LiDAR sensors, paired with a “unique approach to logic,” according to Parker.
The logic works on a hierarchy of rules to allow for flexible and efficient driving, which resembles human judgement. So the vehicle is programmed to ignore “low-priority” rules of the road if required, such as speeding up and abandoning lane discipline in order to safely overtake a vehicle or obstacle.
Solving first world problems
Chief technology officer Emilio Frazzoli told MIT News in March: “[Singapore is] concerned with reducing congestion and trying to get the public to use public transportation as much as possible. At the same time, they’re facing an increase in population by 30 percent within 20 years or so. You cannot keep buying more buses and digging more subway lines. [Singapore] sees [driverless taxis] as a critical vision for their future.”
Parker tells Techworld that the company wants to “fulfil rides much quicker.”
“The driver isn’t exactly going to cancel on you,” he says. “The driver won’t get lost or take the next exit from the freeway. These are pretty first world problems but I have found how much more comfortable Uber was from taxis, and we want this to be the next step in comfort.”
And what about Uber and its own work with autonomous vehicles?
“Uber is an amazing competitor because they are moving fast and have a lot of money,” says Parker. “I think we have some very unique approaches to logic that are hard to copy. This will be proved out over the next couple of years, as they come to the roads, as to who has a robust solution and who has a solution that is still rough around the edges.”
Going for a ride
Techworld went for a ride in the nuTonomy car around the One North tech campus in Singapore this week. The car was pre-programmed to follow a specified route. A safety driver was sat in the front seat - but once the car was set to autonomous mode he didn't touch any controls.
That morning some new construction work had started right in the middle of the pre-planned route, so the car had to adapt on the fly, approaching the traffic cones slowly before jerking onto an alternative route around the obstacle, with the construction workers looking on sceptically.
The rest of the ride was pretty impressive. The car can be switched from manual to autonomous using a button on the dashboard and will then follow a route - albeit jerking a little into turns and with a fairly heavy touch on the brakes. The engineer explained that the jerkiness was a symptom of the safety requirements of the car as it drives cautiously and gives obstacles a wide berth, and that these will eventually be ironed out.
The car’s cameras identify pedestrians and obstacles. On our ride the cameras spotted a pedestrian having a cigarette on the pavement and slowed down in case they stepped out. The front-mounted camera read traffic lights to know when to stop or go. The engineers did admit that the cameras were having some trouble identifying the difference between small children and traffic cones.
Parker says nuTonomy is aiming to deploy “thousands” of driverless taxis across Singapore by 2019.
It is also in conversation with partners in the automotive industry right now and would like to see it taken on by ride-sharing apps too.
“We think that’s what happens first,” says Parker when speaking about ride sharing apps such as Singapore-based car hailing startup Grab. “We can constrain where the car goes, when it operates, we can update the software. We have a lot more control.”
Speaking more generally about the automotive industry, Parker isn't phased by bigger competitors like Elon Musk's Tesla, saying: "They're an amazing company and power to them, but we are focusing on the logic."
"The OEMs (Original Equipment Manufacturers) have to be careful right now. If I put it on a Mercedes, it has to work on every road. That's a big bar. It's smaller to just try this neighbourhood and be the taxi for this neighbourhood."