Doubt. Lidar can make 2 class cars better by being level 2, perhaps most companies will be ready to certify cars on level 3. But 99% of complex work (i.e. AI production) will still rely on camera data at level 4. / 5 (driving vehicles only )
Let’s look at the facts here: LIDAR provides accurate distance data. That’s all it does. There is nowhere near that which makes self-driving cars difficult to eliminate.
The pedestrian stands at a corner. Do they cross this road or another road? Cameras will need to know the target, its shape, movement, and AI to determine its objectives.
LIDAR will tell you that it is 1.7m tall and 7.32m away.
The traffic light is out of control. A traffic officer shakes his hand and points to his other hand. Is he telling you to stop and the other one to go, or is he telling you to go away? Then 100% camera data is AI.
LIDAR tells you that there are 1.5m and 8m traffic lights above the ground at the intersection. LIDAR tells you that there is a 1.8m long object in the middle of the intersection.
Just ask yourself this: If LIDAR was free, how can we stay close to self-driving? Like, use 10 LIDAR sensors on the front, 10 on the back, 10 on each side, placing them in the ears and cups on everything I care about. They are free. How can we stay close to self-driving? Answer: no change at all. Almost every effort and challenge is required to teach AI to know more about video. LIDAR is an additive, such as ultrasound or radar.
Source link
