Google's new patent: Autopilot can predict the behavior of the car

Google has recently filed a new patent application for its self-driving car, which can prejudge the behavior of other cars on the road, thus avoiding accidents such as rear-end collisions. Determining whether the front car will turn or slow down at the next intersection is the basic skill of the human driver and the key to ensuring that there is no rear-end collision. In addition, it is the driver's "second instinct" to look ahead when driving. For unmanned cars, these basic capabilities remain a challenge.

But Google recently filed a patent application for its self-driving car that detects and tracks the brakes and indicator lights of other vehicles on the road.

Through the front camera, this technology allows Google's driverless cars to predict the state of motion of other cars. At the same time, the on-board computer detects the change in the brightness of the car light to determine whether the driver turns on the indicator light.

From a computer perspective, Google will use a set of algorithms to identify the pixels of the image in the camera and analyze its color to determine if there is a pulse inside. If there is a pulse, the algorithm will determine how the driver might operate the indicator and then feed the information back to the central system.

If such a patented technology can eventually develop into an industry standard, it will have broad application potential. Google said in the patent application that the vehicle in front could be a car, truck, motorcycle or bus.

Nonetheless, since Google has acquired a large number of patents related to self-driving cars, it is still unclear whether the newly applied technology will eventually be applied to its autonomous vehicles.