分享一位前激光雷达支持者近期转变想法的视角

Group 1 - The article discusses three main perspectives on the use of LiDAR in autonomous driving: complete avoidance, safety redundancy, and deep integration into driving systems [1][3] - The pure vision camp argues that human drivers can achieve high safety levels without LiDAR, while the safety redundancy camp believes LiDAR can enhance active safety but is not essential for driving [4][6] - A shift in opinion is noted, with some individuals moving from the safety redundancy camp to the pure vision camp, emphasizing that current technology allows pure vision to achieve the safety benefits of LiDAR without its drawbacks [4][5] Group 2 - LiDAR is seen as beneficial for obstacle detection but has limitations in understanding the physical world, which can lead to inefficiencies and delays in processing [5][6] - The article highlights the importance of improving computational platforms to enhance the responsiveness of autonomous driving systems, with a focus on reducing the perception-to-execution time [7] - The potential for future development of LiDAR is acknowledged, but it is stated that the current technology landscape favors a combination of cameras and microphones for perception [5][6]