You can see around corners without relying on advanced optical equipment

[Erica Zhu Feilong Jiangli]

Chongqing Feilong Jiangli collects the latest information for you, so that you can get the most cutting-edge technical information. With the development of autopilot technology in the direction of wider application, it has made a lot of progress in imaging. However, a technical challenge associated with improving safety is the ability of a self driving vehicle to not have a strong ability to observe the surrounding corners.

Imaging and sensing are the key technologies of automatic driving, and also the key areas of 3D laser scanning system, such as lidar (optical detection and ranging). Such a system allows automatic driving vehicles to detect and avoid obstacles by using rotating laser beams to drive safely. However, the current imaging technology needs to be improved in order to improve the performance of the self driving vehicle. According to foreign media reports, researchers at Boston University have conducted a study to develop a new method that allows artificial intelligence components in autonomous driving vehicles to see the corners of vehicles around the University.

Researchers at Boston University have succeeded in making autonomous vehicles perceive the surrounding corners without using advanced optical devices. The technology is based on a standard digital camera and a special algorithm called "computational periscope".

It works like a toy periscope, a device consisting of two mirrors or prisms that allow people to see objects blocked by obstacles. Although the technology does not rely on mirroring, the principle is similar. The algorithm developed by Boston University uses the fact that light can be reflected from structures similar to walls in different modes to assess the degree of disorder created by different objects. The main task of the AI component is to interpret the scattered reflection image.