Comment Re: As far as the "gaping pothole" goes... (Score 1) 289
If they are unrelated arguments the why did you puth them in the same paragraph? If they are unrelated then your following statement is unsupported; Recognizer parking marks: already done some years ago, it's basic computer vision technology.
So the second part has nothing to do with parking so why is it in the thread about parking?
Here is how the Audi did it.
You said
"References? If you are talking about the DARPA Grand Challenges [wikipedia.org] noe of the winning technology was even close to commercially viable. They were proof of concept at best."
I was answering that.
The car uses an array of internal and external sensors to get its position: Audi claims they can be as accurate up to 10cm, but only if they have access to special laser sensors inside the parking structure (four of those scanners had been set up in the parking structure to support the demo). These might be redundant in the future, as the car maker is working on a laser sensor that will be integrated in the car itself (think the sensor tower on top of Google's self-driving car, but completely integrated in the chassis).
The self-parking system also needs access to the car park's management system, in order to find and allocate a free parking space and transmit the route to the car. Since most modern car parks have more than one level or are underground, GPS-based positioning is not really an option, so instead the management system uses Wi-Fi to transmit the route.
That's lots of infrastructure. In the second video the driver selected the spot and there was a car to park next to. The Volvo scenario has three strategically placed cars to mark the parking spot. Notice that the car drove across many painted lines, a no-no in most lots, and ignored many parking spots. That video looks very suspicious considering the car starts and ends in the exact same place every time. Running a car on a script with a simple algorithm to stop and wait for an obstacle to move is a trick.
I doubt that any of those vehicles given an open parking lot would know where to park. The forward parking problem is not "solved".
https://www.youtube.com/watch?v=Ijk3iXu9Kpk
This clearly shows that CV was able to detect street marks as soon as 2009.
More on this https://www.youtube.com/watch?... https://www.youtube.com/watch?...
It may not be "solved" as in "everybody will have it on every car tomorrow", but as in "the tech is proven, is there and it's evolving steadily".
