 The idea that we could be living in a parallel world is a fascinating, if spooky prospect, spun out of science fiction. But for many who make a living grappling with science fact, it's actually a useful way of viewing the world. Splitting reality into different planes of existence, it turns out, makes designing automated systems like smart home devices and robotics much easier. And a recent study by researchers from China shows that it's leading to important advancements in the design of driverless cars, most notably through the concept of parallel driving. The thinking behind parallel driving goes like this. A driver and her vehicle are, at any given moment, cruising along in two parallel dimensions. There's the familiar physical world, which encompasses all reality that is perceived by the driver and that governs the motion of the car's hardware. And then there's the artificial world, where virtual versions of the car, other cars and the road, are simulated by the car's software. It's in this virtual space that data like speed, road conditions and proximity to surrounding vehicles are continuously gathered, analysed and fed into countless simulations run on the fly. Over time, that data provides an increasingly complex picture of the real world. So complex, in fact, that the car's software can begin to map out the best route for a trip and, to some degree, even predict future road or traffic conditions. In other words, the car learns to become independent from the driver. While researchers are still years away from full automation, the research team from China is making strides in adapting this so-called predictive learning strategy to real vehicles. Their system, dubbed iHorizon, uses the technique to classify drivers according to their driving style, calm, normal or aggressive. That information helps onboard software predict driving speeds and energy use on the way to a selected destination. For the growing number of electric car drivers, that means less uncertainty about when and where their battery will run empty. Though it might involve a strange way of looking at the world, parallel driving offers researchers a valuable tool for automating the driving experience. And, ultimately, it could ensure that driverless cars run as smoothly, safely and efficiently as possible.