Hacker News Reader

Comment by rwmj | Hacker News Reader

The BBC had a video of the journalist travelling around SF in the completely driverless Cruise cars: https://www.bbc.co.uk/news/av/technology-63077437

They seem to be doing alright for a technology which is supposedly "going nowhere". (Yes I know these cars sometimes stop, but that's better than having them crash).

View on HN

There is a problem that the BBC won't discuss. LIDAR is very expensive to put in a car and is looks very ugly to have that contraption on top of it. A double negative to sell this to customers and it only works and is trained in SF so far.

To counter the expensive point(s) that LIDAR critics have mentioned you then have Tesla's Fools Self Driving (FSD) demoware, which still doesn't work and even if it does it works worse than LIDAR, hence why Elon's 2020 robo-taxi prediction was completely off especially even if you gave him more time.

But at least Lidar works better here than vision, only for this robo-taxi use-case here and not for selling such cars with expensive LIDAR systems to individuals at scale.

There's nothing fundamental about lidar that means it can't reduce in price. The 20th century is littered with examples of highly complex devices that we made in such high volume that they became cheap. Integrated circuits, optical disk drives, hard drives, all the different display technologies - building ten of these would cost millions, but building 100 million drove the unit cost to almost nothing.

I agree it's possible to make a lidar unit for $25 down-the-line; there's nothing fundamentally expensive. However, it's not just a question of volume. It took decades of Moore's Law before computers became cheap. Optical disk drives took a decade or two as well for reading, and more for writing. Display technologies were a little bit faster, but not fast.

I'm pretty convinced if we wanted to make 100 million lidar units in 2023 or 2024, the cost would be approximately* the same as it is today.

* From a computer science big-O perspective. Simple volume economics might drive the cost in half, or even a quarter, but not down by an order-of-magnitude.

LiDAR is finally getting cheap. OEMs (like VW) are very price sensitive. It is estimated the sensors from Valeo cost about 500 dollars. The fact that you see more and more normal passenger cars with higher resolution LiDARs means that LiDARs are getting cheaper.

The Audi A8 used Valeo's (with Ibeo) first generation low resolution LiDAR Scala 1 from the automotive supplier Valeo. Mercedes new models will be using Valeo's second (or third) generation LiDAR. All these are used for L2/L3 assistants. Valeo is a traditional large automotive supplier.

Luminar, a public company from the US, cooperates with Volvo. Some models will come with a LiDAR in the base configuration. These are "new LiDARs" with high resolution.

Innoviz, a 'startup' from Isreal, will supplies LiDARs to VW. Its angular resolution is (in its focus area) about 0.1 (or 0.2) degrees, which is sufficient for higher levels of autonomy and surpasses/equals the resolution of the expensive Velodyne sensors of the past. They will probably be in the same price range. Due to the limited FOV due to the technology, you will need multiply LiDARs.

Many new models from Chinese car brands will also ve equipped with a LiDAR. Most of them with Chinese LiDAR manufacturers like RoboSense or Hesai. Some are equipped by European manufactures like Ibeo/ZF. For example, there is the automotive sensor AT128 by Hesai. It targets normals vehicles (see price range above) and claims a similar performance (except for FOV, so you need multiple) like the Velodyne Ultra Puck (~$50000).

So costs of LiDARs are a not the very expensive obstacle they were in the past. The only problem could be that the new LiDAR manufactures cannot scale up series production. For example, Ibeo just filed for insolvency because they could not close another round after aggressively increasing spending in the past years.

https://www.youtube.com/watch?v=j9MIYi7b4gc Here is a 1 hour and 45-minute video of a Tesla driving around the bay with zero takeovers. I drive my Audi every day without lidar by using the two cameras and processor built into my head. You don't NEED lidar, but in theory it should make solving the problem easier.

The cost of LIDAR keeps going down. Velodyne is selling a LIDAR sensor for $100.

https://velodynelidar.com/media-coverage/ieee-spectrum-highl...

LiDAR doesn’t work in the rain. But Tesla’s approach is foolish?

Tesla taking longer than Elon thought doesn’t mean they are still not miles ahead… who else is even close? (On a rainy day)

> (Yes I know these cars sometimes stop, but that's better than having them crash)

It kinda depends where and when it stops. If it panics in the middle of an intersection with an incoming car, you get this:

https://www.cnbc.com/2022/09/01/gms-cruise-recalls-and-updat...

These things are not robust or reliable. These are publicity stunts.

Reply on news.ycombinator.com