Hacker News Reader

Comment by ziikutv | Hacker News Reader

Heres a question, I know a lot of companies are investing their time/money into working on self-driving cars.. but from what I see, a lot of the cars use the same sensor. LIDAR and cameras being the main sensors. Of course, we have already seen Tesla perform well with these sensors.

However, my question: Is/Was this industry only held back by sensing?

edit: Restructure paragraph

Regards,

View on HN

For testing purposes, today's clunky rotating LIDAR units are OK. For deployment, something better will be needed. That's coming along well, but due to lack of a market, nobody is producing in volume yet. I'd bet on Continental, the auto parts company, which is using flash LIDAR technology from Advanced Scientific Concepts.[1] That's known to work well, and ASC has been selling it to DoD successfully. Space-X also uses their units for docking the Dragon spacecraft.

Quanergy has been in "fake it til' you make it" mode for a year now; they announced an impressive product 14 months ago and didn't ship. Their web site looks as if they're shipping, but they're not. Now they're trying to get in on the Trump border wall security business.[2]

[1] http://continental-automated-driving.com/Navigation/Enablers... [2] https://www.forbes.com/sites/alanohnsman/2017/03/24/silicon-...

Their logo is a curious mashup.

Tag-line "the future in motion".

Logo character: a horse.

Really?

It's as if they wanted to present themselves both as an old and respectable company (hence the horse, and name), and yet still modern enough to define the future.

The mashup though is so clumsily done it makes me believe the company is either a very early stage startup, or an utter scam.

145 years old, 39 billion in revenue, 212000 employees. Quite the scam :)

https://en.wikipedia.org/wiki/Continental_AG

Let me revise my thoughts...

http://www.conti-online.com/generator/www/start/com/en/index...

Their main website has a splash flash page, which lead me to believe that they didn't put a lot of effort into "public", but not customer, branding. That's reasonable since they're a manufacturer.

However, I've also noted that they're German.

It's not very surprising that a German enterprise isn't spending a lot of effort on its English-language portal.

It's a combination of elements: sensor price, progress in ML, available computational power on board of the vehicle, the advent of electric vehicles, lots of available data and the ability to process all that data, the fact that it is a huge market waiting to be unlocked.

All those together account for a good chunk of the reasons why right now there are at least 10 and probably more parties active (parties that have a credible shot at achieving this).

The real asset is likely all the collected data from the sensors, and not the sensors themselves. The early players have a substantial head start. Their AI software basically has a huge private training corpus that makes it perform better.

Tesla would suffer a bit since their data is less rich without lidar, but make up for some of that with the larger aggregate miles traveled.

Not to say some players don't have better lidar than others, but the sensor quality/price should even out over time since credible third parties are in the business of selling it.

Also, Tesla probably has a plan B for duct-taping LIDAR sensors onto their production cars in case the whole camera + radar + ultrasound approach doesn't work out (or if LIDARs come down in price and turn out to make the system more reliable). With a massive vehicle base and global network of service centers, there should be a possibility there.

Tesla does not use LIDAR, they use radar.

Reply on news.ycombinator.com