4 ms·

There were major problems with every demo I saw of the product. In one video, the camera lost tracking of the road due to lighting conditions not being favorab

There were major problems with every demo I saw of the product. In one video, the camera lost tracking of the road due to lighting conditions not being favorable for contrast. In another, Hotz seemed far too amazed that the car properly took a gentle bend in the highway. Confidence is not something I would assign to his product. Or any product that is a camera-only lane-follower. That Holtz caves at the slightest thought of a regulator looking at his work easily shows how immature he is as an engineer.

vvanders 10y ago
Yup, can't agree more. If you're in the space you had to know these type of questions were coming and you'd have to deal with it. There's a long road from a working prototype to a shipping product.
dexterdog 10y ago
Where was it established that it was camera-only? My understanding was that it used the existing sensors with the camera.
gst 10y ago
> There were major problems with every demo I saw of the product. Doesn't surprise me, after reading http://www.bloomberg.com/features/2015-george-hotz-self-driving-car/ http://www.bloomberg.com/features/2015-george-hotz-self-driv...: "Instead of the hundreds of thousands of lines of code found in other self-driving vehicles, Hotz’s software is based on about 2,000 lines." In general I like the approach of using deep-learning instead of complex hardcoded rules (see the article above for more details), but even with deep-learning you're going to need a complex ruleset in order to deal with the edge-cases. Getting a prototype of a self-driving car working is probably much less than 5% of the overall amount of work that's required. Getting the car respond correctly to the long tail of edge cases is the other 95% of the work.
- smrtinsert 10y ago
  Who brags about 2000 loc when safety is at risk? Seems very immature, and not confidence inspiring.
  - Houshalter 10y ago
    Actually I think fewer lines of code is less risk. Added complexity creates more points of failure, more possible bugs, more unexpected edge cases. Complexity is not inherently desirable.

vvanders 10y ago

Yup, can't agree more. If you're in the space you had to know these type of questions were coming and you'd have to deal with it. There's a long road from a working prototype to a shipping product.

dexterdog 10y ago

Where was it established that it was camera-only? My understanding was that it used the existing sensors with the camera.

gst 10y ago

> There were major problems with every demo I saw of the product. Doesn't surprise me, after reading http://www.bloomberg.com/features/2015-george-hotz-self-driving-car/ http://www.bloomberg.com/features/2015-george-hotz-self-driv...: "Instead of the hundreds of thousands of lines of code found in other self-driving vehicles, Hotz’s software is based on about 2,000 lines." In general I like the approach of using deep-learning instead of complex hardcoded rules (see the article above for more details), but even with deep-learning you're going to need a complex ruleset in order to deal with the edge-cases. Getting a prototype of a self-driving car working is probably much less than 5% of the overall amount of work that's required. Getting the car respond correctly to the long tail of edge cases is the other 95% of the work.

smrtinsert 10y ago
Who brags about 2000 loc when safety is at risk? Seems very immature, and not confidence inspiring.
- Houshalter 10y ago
  Actually I think fewer lines of code is less risk. Added complexity creates more points of failure, more possible bugs, more unexpected edge cases. Complexity is not inherently desirable.