In the “3rd Intelligent Hardware Innovation and Entrepreneurship Interactive Forum Artificial Intelligence and 3D Vision” sponsored by Huaqiang Electronic Network, Shenzhen Sagitar Juchuang Technology Co., Ltd. partner Wang Yanxiang said: “For the development of autonomous driving, the industry generally considers the need for lidar , Millimeter-wave radar, camera and other multi-sensor fusion. Although the camera is not effective in strong light and other environments, it is widely used in automotive ADAS systems due to its high cost performance. The laser pointer radar is currently expensive, but it is necessary for automatic driving. Fewer sensors. ”
He further stated that in autonomous driving, lidar can play functions such as generating high-precision maps, real-time positioning, obstacle classification, dynamic object tracking, and obstacle detection. Specifically, the advantages of laser radar based on vision SLAM map creation and positioning in autonomous driving are stability, small amount of data, and high accuracy in positioning and map creation. The only disadvantage is that the sensors are expensive. Obstacle detection based on lidar has the advantages of not relying on light, positioning first and then identifying, full-view of a single sensor, high spatial positioning accuracy, and small calculation, but also has the disadvantage of low recognition granularity. Therefore, autonomous driving is not only inseparable from the laser pointer radar, but also requires the fusion of multiple sensors and the application of deep learning to improve the level of autonomous driving. But the first issue that needs to be addressed is cost.