I can totally see how a human driver can crash in this situation if he or she is solely focused on the lane markings. This puts the lie to Tesla and Mobileye's claims that camera-based computer vision is enough for autonomy just because humans use two low-res cameras to drive.
> This puts the lie to Tesla and Mobileye's claims that camera-based computer vision is enough for autonomy just because humans use two low-res cameras to drive.
No, it doesn't. It just means that current Tesla's implementation of camera-based computer vision is not yet good enough.
Part of me wonders whether or not we're going to see mutual improvement of infrastructure with self driving cars in the future. This seems like something that you could obviously fix by adding in some kind of passive or active flag that, to a self driving car, says "woah, you're /really/ not supposed to be here." I know we're supposed to be focusing on building systems that are as capable as humans are as drivers. Ho…
We should also not allow software that hits a static object, the car should be able to detect solid object, should be able to detect it's speed relative to the car and to the ground, and it should detect a possible collision and take action.
You can put electronics on the roads but if the cars will depend on those then it will fail when those electronics are missing.
I wonder how easy is it for Tesla to 1) find the exact problem and 2) fix it (provided it is software based). Does it mean that they first have to figure out what some "black box" neural network(s) are getting wrong and then retraining (on simulations), etc?
What if city X decides to repaint all its buses and that confuses the object detection system? How log would it take to get the (hot)fix out of the door...
> This puts the lie to Tesla and Mobileye's claims that camera-based computer vision is enough for autonomy. How are these things related? Are you suggesting that cameras are not sufficient for self driving?
The companies are stating that they won't be using LIDAR or Radar, and that cameras are enough. Cameras tend to rely on markers of some sort - and in cases like the markers on the road were very misleading. Even a short-sighted beginner human driver who was told to carefully follow the lanes would have made the same mistake. And cars running purely on cameras are comparably to short sighted baby human drivers.
At the rate camera technology is advancing, pretty soon human will be the short sighted babies.
I really dislike the marketing around "autopilot." The common defense of Tesla's autopilot is that there are disclaimers, and the driver should always be at 100% attention. Well, look at this, copied from Tesla's website: > Full Self-Driving Hardware on All Cars > All Tesla vehicles produced in our factory, including Model 3, have the hardware needed for full self-driving capability at a safety level substantially gr…
could be a well-meaning tech culture screw up; Most tech people know that airplane autopilots weren't end-to-end automated for the majority of their existence
As a reasonably self-aware dumb tech person, I would say that this defense would not fly with me any more than "any ____ should know that the product can't actually do that." There's a reason that such consumer protections are in place against claims like this, and especially with something like an auto-pilot, I think it's down-right irresponsible to allow marketing to dilute the reality of the performance. This isn't like GB vs GiB for storage, which is annoying but understandable, it's a tool that has very serious consequences if what is advertised doesn't match up with reality. Caveat Emptor only covers so much, and when you have non-stop marketing material about autonomous vehicles and auto-pilot being produced about your product (whether internally produced or produced by third parties), there is a responsibility to set out a clear expectation for the product.
This (video) is another instance where the lanes split apart but the part in the middle has absolutely no markings to indicate that you shouldn't drive there. This (photograph) is what it should look like: http://s0.geograph.org.uk/geophotos/02/24/67/2246762_3a6ea81... Is this (bad line painting) common in the US? Edit: on a second viewing, I think it actually does have some chevrons, but they're very faint.
It's pretty common to have poorly painted roads. There are definitely some roads in my area where the lane lines are no longer discernable at all if it's raining. (Newer roads have embedded reflectors.) If self-driving cars are going to survive, they need to be able to deal with the environment we have, not the utopia where all the road lines stay well painted. ;)
moreover even perfectly painted lines are invisible under a thin coat of snow.
Is it the driver or the car doing the final braking in the video? Even if autopilot followed the wrong line it should perform emergency braking or evasive maneuver when the barrier is right in front of it. This is not just one error, it's two.
To me this is very strange. Why do the road markings trump the objects coming up? Obviously if a giant-ass obstacle is coming up, the car should be prepared to drive around it, and to decelerate as a last resort. Moreover, if the obstacle is moving, the car has to try to extrapolate its movement in order to avoid it. How else can you pass trucks and such? Also why can’t these cars be tested autonomously with soft top…
What a terrible video till.e Also, why are people recording with their phones while driving (or in this case letting drive)? This doesn't look like a dashcam
Fun fact - given the current implementation of Autopilot, it is insane to expect an autonomous vehicle. There should be some kind of common sense/intelligence test for would-be users of Autopilot. So far, every incident involving Autopilot that I have heard about or looked at can be summarised with "Driver was being a dumbass"
Fun fact - If you defend your system as "Driver is responsible for every accident, even if the car drives full speed into a static obstacle", then your system will never be summarized as "at fault".