LIDARS for robocars are everywhere at CES


I'm back from CES and my first report concerns the trends in the LIDAR industry I saw from the 43 LIDAR companies exhibiting there. I talked to most of them. Those trends include lowered cost, more robust instruments and scores of paths to victory. There is also much more attention on LIDAR for the ADAS market. Bosch even said it would make a LIDAR, but said nothing about it.

Read LIDARS for robocars are everywhere at CES on


I really hope Tesla is secretly working on a cheap lidar device that it can cheaply install in its model 3 (and beyond). (Maybe install could be behind the windshield? Not sure if that's feasible.)

Within the next few years it looks like the price is coming down enough for this to be a win. Additionally, AI processing power as well as techniques have advanced so much in the last decade that I think its possible now to use the data in a way that is not the crutch that early self-driving car designs used it as.

For a new company looking to design a self-driving car, one or more less-than-360 degree lidar sensors is probably the way to go. For Tesla it's a more difficult choice, as they'll have to retrofit, but they've got a creative team, and they should be able to figure something out. FSD (which should be feature complete soon even without lidar) costs $7,000 and is scheduled to only go up in price over time. For $1,000 more you can probably increase the quality significantly. And if the lidar data is used correctly (no significant preprocessing prior to feeding it into a neural network), integration into the software that Tesla already has shouldn't be too difficult.

Training will take some time. Maybe that's the rub. How do you deploy it widely enough to get the NN trained, before it's useful?

Okay, plan B: Deploy it as a surprise feature that comes with every Model Y car! (Retrofit to 3s once the network is trained.) Not sure there's enough time before the Ys start rolling out, though.

Is of course only possible for the forward view. Many like LIDAR of some degree in all directions, though it's easier to believe you can survive without it there. Stereo can do a job up close but Tesla does not have stereo cameras in the other directions.

Tesla says "FSD will go up in price" and perhaps it will, but it's not just up to Tesla. Many are working on the systems, and plan to sell to other players. There will be a competitive market and Tesla can't just name any price they like.

is only possible for the forward view because only the forward facing window is called a windshield. But that's also the view that's most crucial. Sure, many teams are going with 360 degree, but I think that's usually just because they got stuck on that design many years ago, and it's hard to change. They used lidar as a crutch and now it's hard to take the crutch away.

Tesla is well positioned to do better than that. Yes, the price of FSD is not just up to Tesla, but the value of FSD will be much much higher than it currently is once it is approved for unmonitored use.

It's unlikely there will be a very competitive market for FSD for a long time. Hopefully I'm wrong about that, but I doubt it. There will be several players, of course. Few of them will be as good as Tesla's, though, and every one of those few will be very expensive. Whether there will be few or several that are legal to use in most places is something the legislatures have yet to decide.

Add new comment