Google’s latest patent aims to help driverless cars better understand humans
We have all seen humorous videos of what a world fully consisting of driverless cars will look like. It’s fascinating as well as somewhat hair-raising seeing cars manoeuvring through swarm intelligence systems and missing each other by inches at high speed.
The latest patent from Google aims to take this a step further, by aiming to help its fleet of driverless cars better understand other cars’ turning signals.
The patent, which can be seen in full here, is for “vision-based indicator signal detection using spatiotemporal filtering”, and was originally filed in June 2013. Google’s driverless cars have been in and out of the news for various crashes over the past few months – all of which the search giant claims is not its fault – so the patent makes sense on that front, although as driverless cars gradually become more of a majority on the road then the technology will not be as frequently required.
Based on the patent’s abstract, Google’s idea is to have a camera on the side of the car which captures other cars’ movement and detects by colour and brightness with a group of pixels the likelihood of another car using its turn signal. Among Google’s other planned inventions is technology that recognises buses and potholes.
Google had a major validation of its operations when the National Highway Traffic Safety Administration (NHTSA) wrote a letter which in effect opened the door for its self-driving cars to be legal on US roads. In November, a research note from IHS Automotive argued the company was the current technological leader in the self-driving car space.