Labeling LiDAR data is extremely time-consuming and costly. You can’t imagine—a few meaningless dots there actually represent a car, and yet many people believe this is ideal for self-driving.

This is a “car,” according to LiDAR.

My vision based robot drives me everywhere every day, it just works. Many of us work in industries that use vision driven systems regularly. We understand the capability, the general population is just ignorant. That’s not an insult to them it’s just a reality of the lack of education of subject matter.

Arguably lidar and radar is good for certain things like exact positions of objects and obstructions, whereas cameras are good for labeling objects in general, I think in 100 years we’ll use all of the sensors but at this moment and with compute limitations we will stick with cameras only, if humans can do it then so can computers, but to achieve significantly better than human status, we might require additional sensors for more specific positional data

Expand the possibilities of your metabolic research. Resipher tracks real-time cellular oxygen consumption in standard 96-well plates, delivering continuous real-time data directly from your incubator. Request a free virtual demo or quote today >>

Like two people believe that. Mark rober and luminar lol

Does @Scale_AI offer efficient LiDAR data labeling methods?

Lidar is a good verification for your camera network.

Still puzzles me they think $1k per car is cheap, and said prices is dropping for LiDAR..

In machine learning world , data-labeling is the most important and time-consuming. But LIDAR Data is hard to label and maintain.

It's ideal for not crashing into stuff. Which is not a problem for Camera's either....

I will go out on a limb and say the only feasible technology is whatever Tesla uses. All other technology is inferior, no matter the use case. Correct cultists?