Researchers beyond the US are keeping pace with Silicon Valley when it comes to autonomous cars. Nicole Kobie reveals the R&D that will see driverless vehicles hit the roads soon.
Google’s autonomous tests have racked up more than two million miles, while Tesla’s Autopilot has clocked over 50 million miles – and a death. Uber-owned Otto just shipped 5,000 bottles of beer through Colorado in a self-driving lorry. Meanwhile, Singapore has rolled out its first autonomous pods and driverless taxis.
It may sound like the rest of the world is falling behind in the race to self-driving cars, but researchers from academia and industry are hot on rivals’ heels with plans to trial autonomous cars and transport.
As many as 100 Volvos will take to London roads in a trial of Nvidia-based technology, alongside similar programmes in Sweden and China. They won’t be driven by researchers, but by “real families”, the company said. The first semi-autonomous cars will hit the streets in early 2017, with the aim to switch to fully self-driving by 2018. Volvo will take responsibility if anything goes wrong.
“It’s an experiment with the technology, but more a social experiment on how people react and respond to it,” said Danny Shapiro, Nvidia’s senior director of automotive.
This isn’t the first time driverless cars have been tested on English roads, but it will be the first on public roads. That’s largely due to legal hurdles. A consultation on “advanced driver assistance systems and automated vehicle technologies” took place from July to September 2016, with the Department for Transport analysing the results. Most expect this to pave the way for legislation, so driverless cars can legally take to the streets.
For the time being, testing is restricted to the private roads of university campuses – and test tracks. Ford, Jaguar Land Rover and Tata Motors met up at the MIRA test track in England in October 2016 to test connected and autonomous cars developed under their Autodrive project, with cars showing off how they communicate to avoid collisions and red lights. “It was a big step forward in demonstrating how the technology has leapt forward,” said Stephen Lynn, head of marketing at the Transport Systems Catapult.
The trials will head to closed-off areas of the UK in late 2017, followed by public road trials a year later.
Those trials are looking at adding autonomy to standard passenger cars, but many researchers expect the first truly self-driving vehicles to be pods akin to the transporters in Singapore.
“There are only so many roads you can build,” said Umar Daraz, director of the Institute for Sustainable Futures at Birmingham City University. “What we’re hoping with the autonomous vehicles is to trial them and show that there’s an opportunity to lay down infrastructure around an autonomous vehicle corridor to help people flow from different parts of the city – rather than lay down trams, have more roads, more congestion, and all the issues around air quality.”
In 2017, Daraz will run two trials for the “Insight” project, with pod-like vehicles that will shuttle people around the city, focusing on transport hubs. The pods are designed to help those who are less able, particularly the blind or those with reduced sight. “People are at home alone; because of their disabilities they can’t access the transport system,” he said. “In some ways, this is a transport system to connect those people, to give them quality of life.” Alongside the autonomous sensors and AI, the pods will feature tablets with tactile surfaces to communicate with blind users via “braille-like shapes”, he added.
Similar trials have run elsewhere, with the Pathfinder pods developed by the University of Oxford’s Robotics Institute slowly rumbling through pedestrian areas or “shared space environments”, said Rebecca Advani, senior technologist at Transport Systems Catapult, and technical lead on the Pathfinder project.
Keeping pace with the US
So there’s no shortage of projects, but can anywhere outside of the US keep up Silicon Valley and Singapore? It isn’t a matter of technology, experts argue, but decisions regarding the rollout appear to be taking the slower road.
“The US is far more advanced in this area, as Google has trialled its auto-vehicles for a while, but the size [of the trial] is pretty small,” said Daraz. “I think what we are looking to do is show the tech works, but then work with the regions to see how they could scale it up into their future transport systems.”
Tony Pipe, deputy director of the Bristol Robotics Laboratory at the University of the West of England, agrees. “My feeling is that in some countries there’s been more of an attitude of ‘let’s see where the technology could go first’ and then worry about the public rollout later,” he said. “In other parts of the world, we’re worrying more about how it will roll out before trying it out.”
The rollout could be a bigger challenge than working out the kinks in the technology. Regulation and insurance issues must be decided, alongside convincing the public to accept automated machines.
Pipe said his own tests suggest the results could be positive. His team tested handing off control between a human driver and the car’s AI, with early results looking as if “people are maybe slightly surprisingly keen to trust technology”. Pipe said even he “became almost blasé” about the technology: “I surprised myself in that I very easily relinquished the responsibility to the vehicle.”
That concerned him: “We really should be introducing this technology slowly, not only because of the newness of the technology but also our attitudes towards it. Early results suggest we might be too eager to trust it, when it may or may not be [ready]”.
When will all this research coalesce into a consumer product? Tesla’s Elon Musk would argue tomorrow, saying that autonomous tech will save more lives than it ends by avoiding traffic collisions, but others ask for patience.
Pipe called for a staged rollout, taking until perhaps 2030 before autonomous cars hit the roads. In the meantime, cars will slowly get more autonomy, taking over during traffic jams or on the motorway.
Advani agreed, predicting that autonomous cars will start with pods used in closed systems, while driver assistance features will creep into passenger cars, eventually building up to full autonomy. “That will be a longer-term progression, but it’s already starting to happen.”
Pipe argued for caution, pointing out that human life is “dearer to our hearts than it would have been when cars were first coming onto roads”. He added: “To quote somebody, caution is the better part of valour. If we’re going to be brave and experimental, we should be cautious, too.”