Google’s driverless cars can’t spot potholes or drive in heavy rain

Guess we’re not going anywhere today, then. Image: public domain.

And it seemed like everything was going so well for Google’s amazing new driverless cars. The first set of prototypes has been tested; they’ve driven over 700,000 miles of US roads; they’ve even created a second generation two-seater car with what looks like a friendly face on the front:

Aww, look at its smiley face.

So you could be forgiven for thinking that Google were on the home straight, and we’d all be chauffeured around in autonomous vehicles before the year was out.

But, it turns out, not so much. The most recent issue of MIT Technology Review has revealed a list of the things the cars can’t yet do, as confirmed by Chris Urmson, director of the Google car team. These range from the mildly problematic  - for example, the cars can’t detect the nature of an obstacle, so would swerve around balls of paper as though they were rocks...

...to the downright concerning. Such as not having been tested in adverse weather conditions such as snow or rain. Or being able to detect open manhole covers or potholes.

Perhaps the most worrying issue, however, is the fact that they still can’t operate on most roads. The cars rely on painstakingly detailed 3D maps, which require multiple visits to streets and analysis by both humans and computers: simply downloading Google Maps won’t cut it. And, since the cars can’t respond to unexpected visual signals, like temporary route changes or new sets of traffic lights, these maps must also be updated constantly. That’s a lot of effort, and so comes at a cost.

Urmson assured the publication, however, that engineers are hard at work on all these issues, and he still hopes the cars will be on roads within the next five years, by the time his 11-year-old son turns 16: “It’s my personal deadline.”

Like what you see? Why not follow CityMetric on Facebook or Twitter. Go on, we're lovely.

 
 
 
 

Why do video games find it so difficult to reflect real cities?

A screenshot from Sleeping Dogs, set in Hong Kong. Image: Ubisoft.

Taking a real city and putting it in a videogame sounds like a great idea. You get all that sense of place, all that rich history to draw on; and every city has its own character, its unique blend of people, politics and culture. Great cities are more distinct to a global audience even than the countries they are situated in. From London and Paris to New York and Hong Kong, these metropolises stand as some of the most significant cultural touchstones for people all over the world.

And yet attempts to put these great edifices into a videogame can often be disappointing. Almost without exception, the best cities in videogames are fictional.

The first set of problems in basing a game in a real world city stem from geography. One example of this is the 2002 game The Getaway, which in spite of its age is perhaps the closest anybody has come to a modern Grand Theft Auto game set in London. (That said, the GTA series did dip into London in 1995’s top-down GTA: London, a dip that has fuelled rumours ever since that the series may one day return).

A screenshot from The Getaway.

The problem The Getaway had, as a GTA style game, was the driving. As a setting for a crime game you can’t really fault London – but you wouldn’t want to drive there. The game captured the absurd traffic congestion and frustrating road layout of the city surprisingly well. Which meant driving wasn’t fun, so neither was the game.

By contrast, the more recent Driver: San Francisco is a game that involves, unsurprisingly, driving in San Francisco. This is perhaps the best example of a game nailing down what makes a city work as a location for the type of play the game is offering. San Francisco is the best city in the world for car chases because it’s got those ludicrously cool hills. And from a game design perspective, everything else is a footnote.

Further problems relate to how a game and its characters treat the city and its people – and it is here that scope exists for a game to become horribly unstuck. Consider 2014’s Watch_Dogs, a game about a vigilante hacker set in Chicago.

A screenshot from Watch_dogs.

Watch_Dogs aimed for a fairly naturalistic tone, painting the city as being full of interesting yet familiar, believable characters. This would be fine, except that your principle methods of interacting with these citizens, in your role as their self-appointed guardian, is to hack into their personal information, rob their bank accounts or kill them with near total impunity. It feels very off.

For this kind of interaction with the general population to be entertaining it requires a completely different tone. This is something that recent GTA games have mastered – that cruel, overt humour, the deep sense of misanthropy and cynicism. There are almost no good people in those worlds, so everybody is fair game. Notably the GTA games all take place (with the exception of the aforementioned GTA: London) in fictional cities which draw from the very worst qualities of their real world counterparts: an evil twin of New York, or an evil twin of Los Angeles.

A game that had a much better angle on how the character interacts with people in a real city is 2012’s Sleeping Dogs. Set in Hong Kong, the game takes its inspiration very much from the movie making tradition there. There are corrupt cops, powerful crime families and a heavy emphasis on using Kung Fu to solve your day to day problems rather than guns.

A screenshot from Sleeping Dogs.

In Sleeping Dogs you played an undercover cop and were discouraged from attacking civilians. You could pick on rival criminals if you wanted to, but the game moved away from the random acts of violence that characterise GTA and Watch_Dogs.

This was an approach also adopted by LA Noire, a 2011 game that also leaned heavily on the cinematic heritage of its location, and which also saw you play a cop – albeit with a heavier emphasis on detective work than on kicking people in the face over and over again. This again is a game that sacrificed the freewheeling fun of random violence in favour of a narrower focus that fit more comfortably with the setting.

By tapping into the culture of these cities, both games are trading on familiar and accepted themes. You don’t play Sleeping Dogs and think Hong Kong is a city with a massive organised crime problem: you play it and think it looks like a fascinating place to visit, and that maybe you should watch Infernal Affairs and Hard Boiled again.

The lesson here is that, if you want to embrace the culture of a city within a game, you have to do so with a degree of affection and respect. If I want to feel a connection to a place in a game, it helps if I’m not being encouraged to brutalise the citizenry and massacre its law enforcement officials for trying to stop me.