For the first time, the technology to detect objects outside the field of view has reached high resolution. you can even read the signals in the corner.

Autopilots and other robotic vehicles learn to assess traffic situations better and more deeply than a human driver. This is also helped by technologies to detect objects that are out of sight. However, there hasn’t yet been a way to get a high-resolution image out of the many technologies for rendering objects in the corner. Scientists at the California Institute of Technology have filled this gap by offering a new UNCOVER technology.

Image source: Caltech

Traditionally, you can look “around the corner” if you look at light reflected from other objects, such as walls, in your line of sight. Because the light from such objects is scattered, a more or less complete image can be assembled using AI algorithms. But if there is a point source of light near the desired target – a guide star (guide star), then the hidden object can be observed in much more detail.

Light from a point source allows you to calculate the light waves reflected from the walls and use this information to calculate the light flux of the observed hidden object. The problem is that under normal traffic conditions it is impossible to highlight hidden objects. Otherwise, every pedestrian and every car on the street would have to be given its own “guiding star”.

UNCOVER technology allows you to use a hidden object as a guide star. During the experiment, scientists demonstrated the ability to distinguish complex signs in the form of an asterisk, geometric figures and letters out of sight. The proposed technology, scientists believe, will help improve autopilots on city streets and rovers on distant planets. The rover will not need to cross massive obstacles to inspect objects hidden behind them. If there is light around, it will get an image out of view.

Related posts

Leave a Comment