Text /BY

"Lidar is so stupid, whoever relies on it will be finished." Musk has used this Diss lidar on many occasions. In his view, as long as the computer vision technology is mature enough, automatic driving can be realized only by the camera. In fact, not only Musk, but also many autonomous driving companies have the same idea.

at this CES conference, Intel showed a self-driving video developed by Mobileye, which caused an alarm at the scene. The particularity of this autopilot demonstration is that the demonstration vehicle is only loaded with 12 cameras and no radar is used! By carrying two Mobileye's latest eyes? Q5 chip, demonstrating the vehicle's automatic driving ability without losing Waymo, including difficult actions such as crossing a complex intersection without traffic lights, automatically planning roads and automatically avoiding pedestrians.

Mobileye said that its technical principle is to generate a 3D model through deep learning processing of 2D images taken by multiple cameras, thus providing the required environmental information for autonomous driving decision. In other words, Mobileye has achieved the same function of lidar through low-cost camera and software technology.

In fact, if the algorithm and computing power are strong enough, it is possible to "upgrade" the 2D images output by multiple cameras into 3D images. After all, that's what people do. Through the powerful computing power of the human brain, people can "shoot" 2D images from two eyes to generate 3D images. People can do it, and machines can do it one day. It is only a matter of time.

in addition to making the camera realize the 3D image of lidar, Mobileye also used the camera to build high-precision maps at this CES. Mobileye has signed a cooperation agreement with SAIC, and will? Mobileye? REM, a road network asset management technology, is applied to automobiles produced by SAIC.

this technology can enable vehicles to collect road data through their own cameras during driving, and these data will be encrypted and uploaded to the cloud, which will eventually form a crowdsourcing "road book" with updated facts, which will feed back the cars equipped with the system and improve their driving assistance. This is similar to the high-precision map. At this stage, the generation of high-precision map mainly collects data through the fleet equipped with laser radar, and subcontracting is also the future plan that many high-precision map manufacturers look forward to.

Mobileye said that the cooperation with SAIC will create opportunities for? L2+? And the road network management system used by higher-level self-driving cars. In addition, the surveying and mapping scheme is an open platform, which will provide support for the layout of other car factories in domestic road network management related fields. SAIC will work with? Mobileye? Other partners of global car companies * * * collect road data together, so as to make real-time and high-precision global maps possible.

For L2+ assisted driving and automatic driving above L4, the cost will be one of the biggest obstacles to popularization. An algorithm based on Mobileye makes up for the shortage of sensors, and even obtains the cooperative benefit scheme of autonomous driving through open cooperation, which will effectively reduce the cost of autonomous driving and accelerate the popularization of high-level assisted driving and autonomous driving. Accelerating the popularization of autonomous driving is one of the themes of this year's CES. In addition to Mobileye, other manufacturers also have different directions to explore. Please continue to look forward to Autolab at CES. Other reports brought by 22!

This article comes from the author of Chejia, car home, and does not represent car home's standpoint.