For all the advances in driverless cars that Google has been making, the full market release of the tech giant's automated vehicle is still a long way off, according to project's head, Chris Urmson.
Speaking to the MIT Technology Review, Urmson, director of the Google Car team, highlighted critical flaws in the way the vehicles operate; admitting that the cars are not able to navigate 99 per cent of the US mainland, despite having covered over 700,000 miles in development so far.
His main criticism came from the system's heavy reliance on maps, with each journey effectively requiring programming in. Once following a particular route, the Google car cannot deviate of its own accord, meaning it would be confused by road works and diversions.
They're stumped by heavy precipitation, too, confusing snowflakes and large raindrops that pass in front of its sensors as obstacles in the road. Strong sunlight also poses a particular problem, with the vehicles confusing it for traffic signals.
And if you live on a heavily potholed road (your average British suburban street, then), expect to be regularly forking out for suspension repairs, as currently, Google's car will drive straight into them.
Mr Urmson said: "I could construct a construction zone that could befuddle the car."
At least pedestrians are safe, right? Well, yes and no. While the Google car can detect humans in the road as a solid block of pixels, and act accordingly, it fails to detect those on the side of the road – e.g. a policeman signaling traffic to stop – and would continue on its route.
Despite these quite fundamental flaws, Urmson remains confident that the self-driving car will be fully functional within five years, though he admits that there is still some way to go.
Would you travel in a self-driving car? Have your say below.