Tesla has been pushing camera-only autonomous driving (ditching LIDAR) since well before the chip shortage. I would assume this is part of the same drive.
I'm pulling this from archive.org, because Tesla has removed this blogpost from their servers.
As recently as March, they kept this pro-Radar blogpost up. Only now are they purging this data from their archives. Fortunately, the Internet Archive remembers the history, so they won't find it so easy to rewrite history in their favor.
Sometime between March 2021 and May 2021 (today), Tesla has deleted this blogpost. The above link is now a dead-link. I've included the original blogpost from archive.org in my earlier post, so that you can see the original content.
My expectation is for Tesla to be honest about their history, and not be ones who delete inconvenient blogposts years later. This selective picking-and-choosing of historical posts is immediately suspect, and extremely damaging to the reputation of the Tesla blog.
In any case, the _timing_ of this deletion event tells us everything. Tesla only recently began thinking about Tesla Vision seriously, which provides evidence that this is a temporary supply chain issue, as opposed to a forward looking technological innovation.
Do you really expect, or even want, companies to go back and purge all old engineering blogs which they now disagree with? By calling out an old post as evidence that they are "pro-lidar", you're the one forcing companies to try to curate the public history of their development process.
It’s radar, not lidar, and the point is that Tesla thought the post was contradictory enough to remove it, while all the other content from 2016 is still there.
They haven’t announced anything like that - they’re just using LiDAR rigs to validate data from their vision based approach, particularly distance. It actually even mentions something about it at the bottom of the article you referenced.
No public announcement. They merely have some lidar equipment for testing. I'd be shocked if they haven't already had lidar units for years. After all, you can't compare your approach to lidar if you don't have a lidar unit to compare to.
This might be for improving their camera-based vision system by collecting both camera inputs and accurate ground-truth data via LIDAR to train the system with.
