How did driverless cars cope in last week’s London’s snow?

The pod, in better weather. Image: GATEway Project.

There’s one London driver that didn’t react badly to last week’s snow: Greenwich’s driverless pod.

The GATEway project runs a pod – the same vehicle used at Heathrow to ferry drivers from the parking lot to the terminals – around a 3.4km route on the Greenwich peninsula. The aim isn’t currently to test the automated technology, but to understand public perception to driverless vehicles.

Still, when you have a chance to test driverless tech in snow, you take it, says Jim Hutchinson, CEO at Fusion Technology, the company behind the automated pod tech. “We want it to go beyond what a human driver could reasonably cope with,” he explains. “That is our aim, but right now, for this trial it’s not a requirement that it can do that. It was more an opportunistic thing really: there was snow on the ground, so let’s take it out and see if it’s as good as it should be.”

And it worked better than expected – though there were differences in the automated pods’ winter driving, compared to the usual weather. Hutchinson said he noticed that it drove a slightly different line than usual, off by “centimetres rather than metres”.

“There were some positioning inaccuracies – they weren’t terrible, but the precision was reduced,” he says. “The degradation wasn’t that appreciable, so that was very good from our point of view.”

That’s down to the car “seeing” less well in the snow. “When you coat everything in white, you no longer have the same definition,” he notes. 

This driverless system doesn’t rely on a single type of computer vision. Driving would be more difficult in snowy conditions if it only used LIDAR – “Light Detection and Ranging”, which pings light at surfaces and measures how long it takes to come back.

But the GATEway technology also uses radar and cameras to see. “Because our system uses a broad range of sensors, we’re perhaps a little bit more robust than some systems that are solely reliant on LIDAR, which can be susceptible to more adverse conditions,” Hutchinson says. “By using radar and to some extent cameras as well we can get around some of those problems better than if you rely on a single sensor type.”

The GATEway pods have fewer challenges than full-blown driverless cars that will take to the roads, however. The pods follow a set route on a pathway, so they needn’t worry about missing a snow-covered stop sign.


Driverless cars that do take to roads will need more robust systems for getting around when snow has obliterated road markings and wiped out signage. In response, Ford is testing detailed 3D maps, while Finnish researchers in December sent their driverless car hurtling down a snow-covered motorway at 25mph using radar, cameras, and GPS – but the road was also kitted out with sensors to help with positioning. 

Seeing in a blizzard is only part of the problem of winter driving, of course, and radar isn’t much help if you’re spinning your wheels or sliding off the road. The GATEway pods don’t have traction control featured in most regular modern cars, but sensors help them understand when there’s a mismatch between what the wheels are doing and how fast the vehicle is actually moving. (This is a feat not managed by my neighbour, who spun his tires for ten minutes before trying de-icing spray on the road.)

Faced with slippery pathways, the Greenwich pod did what a “responsible driver” would do, Hutchinson said. It simply slowed down.

GATEway safety stewards (that is, people) were riding in the driverless cars in the snow – Hutchinson admitted they were initially nervous, but eventually relaxed – but no other passengers faced the Beast from the East from inside an autonomous pod. Was he worried about safety? “My biggest concern would have been that the ground was quite slippery – so just people getting in and out of it would have been my biggest concern, rather than people being in it,” he says.

Of course, those left to walk home on the slippery pavements rather than catch a ride in a covered pod may disagree. Indeed, offering rides in inclement weather is the aim. “It’s really going to help in those sorts of conditions, that’s when you want them…  When it’s really unpleasant, it can make a big difference to your day if you have a comfortable means to get to your destination.”

That’s a lesson learned by Hutchinson last week, after his journey home via stalled trains took 26 hours. But unless you live along the pods’ limited route in Greenwich, the rest of us have many more winters ahead of trudging through the snow on foot before this tech is ready to widespread enough to carry us home.

Want more of this stuff? Follow CityMetric on Twitter or Facebook.

 
 
 
 

In many ways, smart cities are really very dumb

Rio de Janeiro’s control centre. Image: Getty.

It’s not news that anything and everything is increasingly being prefaced with “smart”: phones, watches, homes, fridges, and even water (yes, smartwater exists). And it’s not unintentional either. 

Marketeers know that we, the public, are often stupid enough to believe that thanks to their technology, life is better now than it was way back in, say, the primitive Nineties. Imagine having to, like a Neanderthal, remember how to spell words without an autocorrecting algorithm, or open the fridge door to check if you’d run out of milk, or, worse still, interact with actual people.

So it’s hardly surprising that we’re now also witnessing the rise of the so-called “smart cities”; a concept which presupposes that cities that are not technologically  “smart” are dumb, which, as anyone interested in the millennia-old history of cities — from the crypto-currency grain storage algorythms of ancient Mesopotamia to the complex waste infrastructure of ancient Rome, to London’s public transport infrastructure — will know, is not true.

Deployed in these smart cities are cameras and other networked information-gathering devices, load cells and other “sensing devices” detecting passing pedestrians and vehicles, audio surveillance devices listening for gunshots – and even vending machines equipped with biometric sensors to recognise your face. This is not to mention beacon technology — tiny anonymous looking black boxes hidden in trees and on lampposts — which transmits advertising, offers and other information directly to smart phones in the vicinity. 

If that doesn’t seem sinister enough, take, for example, Rio de Janeiro, where, in 2014, the International Business Machines Corporation designed a mammoth “control centre” that integrates data from 30 agencies for the city’s police. 

Described by the Guardian as having “the functionality of a Bond villian’s techno lair”, the then local mayor, Eduardo Paes, claimed the centre was making the city safer while using technology to deploy its “special” police unit to carry out the state’s “pacification programme”. Launched in 2008, the programme, which aims to push out drug gangs from Rio’s favelas, has been criticised by Amnesty International: “in January and February 2017 in Rio de Janeiro alone, at least 182 people were killed during police operations in marginalized neighbourhoods (favelas) – a 78 per cent increase in comparison to the same period in 2016”.

Sinister or not, as smart cities grow, they create new problems. For example, as urbanist Adam Greenfield writes in Radical Technologies: The Design of Everyday Life, neither the algorithms nor their designers are subject to the ordinary processes of democratic accountability – a problem that international academics are currently attempting to tackle.  


“We need to understand that the authorship of an algorithm intended to guide the distribution of civic resources is itself an inherently political act,” writes Greenfield. “The architects of the smart city have utterly failed to reckon with the reality of power.”

The Real Smart Cities project, founded by Dr Gerald Moore, Dr Noel Fitzpatrick and Professor Bernard Stiegler, is investigating the ways in which so-called “smart city” technologies present a threat to democracy and citizenship, and how digital tools might be used create new forms of community participation.

Fitzpatrick is critical of current discourses around smart cities, which he says “tend to be technical fixes, where technology is presented as a means to solve the problems of the city.” The philosophy underpinning the project is “that technologies function as forms of pharmacology”, he adds, meaning that they can be both positive and negative. “The addictive negative effects are being felt at an individual and collective level.” 

An example of this lies in the way that many of these smart cities replace human workers with disembodied voices — “Alexa we need more toilet roll” — like those used to control the Amazon Echo listening device — the high priestess of smart home. These disembodied voices travel at the speed of light to cavernous, so-called “fulfilment centres”, where an invisible workforce are called into action by our buy-it-now, one-click impulse commands; moving robotically down seemingly endless aisles of algorithmically organised products arranged according to purchase preferences the like of which we never knew we had — someone who buys a crime novel might be more likely to go on and buy cat food, a wireless router, a teapot and a screwdriver. 

Oh to be the archeologists of the future who while digging through mounds of silicon dust happen upon these vast repositories of disembodies voices. That the digital is inherently material and the binary of virtual/real does not hold — there is no cyberspace, just space. Space that is being increasingly populated by technologies that want to watch you, listen to you, get to know you and sense your presence.

One project looking to solve some of the problems of smart cities is that of the development of a “clinic of contribution” within Pleine Commune in greater Paris (an area where one in three live in poverty).This attempts to deal with issues of communication between parents and children where the widespread use of smartphones as parental devices from infancy is having effects on the attention of young children and on the communicative abilities between parents and children. 

This in turn forms part of a wider project in the area that Stiegler describes as “installing a true urban intelligence”, which moves beyond what he sees as the bankrupt idea of smart cities. The aim is to create a “contributory income” in the area that responds to the loss of salaried jobs due to automation and the growth and spread of digitisation. 

The idea being that an income could be paid to residents, on the condition that they perform a service to society. This, if you are unemployed, living in poverty and urban deprivation, sounds like quite a simple and smart idea to try and solve some of the dumb effcts of the digital technology that's implemented in cities under the ideology of being “smart”.