-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Thank you for not hitting a person, not only for their sake and for yours, but for the reputation and advancement of autonomous vehicles. Had you been unfortunate enough to seriously or fatally injure that person, we’d have all learned of it through the news.
Glamisduner
Active Member
Looks like the only thing that saved him, was him jumping out of the way. He deserved to get hit pulling a stupid stunt like that. I'm glad you didn't hit him though.
I agree....no warning? the guy would have been hit if he had not stepped back.
Also AP is not the issue. FCW and AEB should be.
The technology seems to have real difficulty with stationary objects. There have been reports of AEB not working for a stationary car for example. Oncoming headlights are also a big issue. Expectations are way too high in my opinion for what this technology can do.
Regarding the sonic sensors, the manual warns that those can't even detect pedestrians reliably when parking so don't expect they can while driving.
Regarding the sonic sensors, the manual warns that those can't even detect pedestrians reliably when parking so don't expect they can while driving.
Standing in the middle of the road and not even looking to see if a car was coming. Way more optimistic than I am.
Tesla has been shown to have some pedestrian awareness. Maybe just at lower speeds. Also quite possible AEB would have reacted at some too-late point, but the driver was already on the brakes and disabling it. Good driving!
Tesla has been shown to have some pedestrian awareness. Maybe just at lower speeds. Also quite possible AEB would have reacted at some too-late point, but the driver was already on the brakes and disabling it. Good driving!
All the above comments, standing in the middle of a dark, rainy road in dark clothing makes the spontaneity of this video suspicious. The OP has been here a long time and he says it was his. My car recognizes pedestrians and stops for them. This is one of the difficult situations as false emergency brakings are very bad news. The emergency braking is mostly radar controlled with increasing video input but in this situation video input would have been almost non-existent. There will be some situations that perfect autonomous driving will not handle, namely suicides which this incident comes close to.
Tam
Well-Known Member
Just like the Uber's case. Initial blame was on jaywalker. When the police viewed the forward looking video, they said it's unavoidable because "she came from nowhere."
Not until when they saw the interior video showing the driver was looking down watching a TV show on the cell phone, that's when they stopped blaming on the jaywalker.
Actually, I would expect the standard safety feature of the vehicle to pick this person up and emergency stop. Other vehicles it there do this with very good precision. Really a sobering moment for the current state of the Tesla software.
hill
high fiber member
Kinda says reams about how distant AP3/self driving still is.
.
2cool4school
Member
I don't use autopilot, period. It has failed me once - could not read the lane lines (I presume), and simply turned itself off at a curve, resulting in my car going into the oncoming traffic. If it wasn't for my quick reflexes, I'd be not as alive as I am right now. 
As much as I agree with the fact that the person standing in the middle of the road was an utter idiot, I also know that the technology to avoid pedestrians IS already used in many other cars. I currently own one of the two cars that I had the occasion of experiencing pedestrian avoidance. My car shows warning in big red letters accompanied by a ding on the dash when it detects a pedestrian walking in a direction that intersects the car projected path. Usually the pedestrian has to be close to the road or on it. I had alerts of people running on the sidewalk towards my car and stopping at the last moment at the edge of the sidewalk.
I think Tesla still needs to do some work on this> it's not about innovating or inventing at this point. I suspect that it may have to do with my buddy Elon's stubbornness not to have Lidars.
I think Tesla still needs to do some work on this> it's not about innovating or inventing at this point. I suspect that it may have to do with my buddy Elon's stubbornness not to have Lidars.
Would an average driver be able to avoid the pedestrian? I couldn't even see her until the last second in the video. The visibility is so low that I don't expect the camera to pick it up. Remember that AP doesn't need to be the best driver, it only needs to be better than the average before it is road worthy. It will only get better over time where as the average driver will remain the same.
5_+JqckQttqck
Active Member
For what it's worth, the videos from the front dash cam does not reflect the actual lighting conditions the driver sees. The fact is, the driver did see that person and chose to test AP capabilities or was expecting AP to do something. They took over when the car was not slowing down nor did FCW and AEB engage.
It's BETA and designed for freeways. Regular road use has it's pros and cons as there are way more variables. Pro, you'll never rear end another vehicle - Con, pedestrians jaywalking are idiots and are not detected by the radar.
That's why it's only Level 2 AI driver assist.
AlanSubie4Life
Efficiency Obsessed Member
The results of the poll here are sobering! I mean, I can certainly understand TRYING autopilot on secondary roads (not highways, not controlled access), under 100% driver control, with extra vigilance (to make sure it doesn’t steer into opposing traffic for the brief period of use). I’ve done that myself out of curiosity. But to USE autopilot as a feature in that scenario? I am shocked that 70% of people are using a beta feature in an unapproved scenario.
And then I find it kind of funny, in a very sad way, that there are people that expect the car to detect this sort of thing. Why do they think this? The car is not magic... No matter the detection system or manufacturer, any car on the road today can encounter situations where it will fail to stop or even react at all. This is just a really hard problem.
When combined with an attentive driver who is not relying on the assist in any way (i.e. behaving as if they do not exist), these systems (AEB, FCA, not including AP which is not relevant here) will be better than the alternative of just a driver with no aids. But at the moment, that is as far as it goes.
EDIT: if AP (or AEB) cannot react (when it is being misused) to a stationary fire truck, it seems unlikely it would react every time to a stationary pedestrian! Reacting to stationary objects I imagine is pretty tough since a lot of things (like the road if it is undulating) probably look stationary to a computer and neural net. Human brains are pretty impressive. If the pedestrian had been moving maybe it would have reacted...or not... Personally, I would not expect AP to stop me if I approached an unexpected traffic jam at freeway speed, if TACC was not following any one prior to the required stop. Have not had occasion to test it thankfully. It might work (I certainly get FCW alerts sometimes so it might help...but when that goes off (even on “early”) it is getting close to a crisis), but I think the chance of someone piling into me from behind subsequently would be higher than if I stopped myself, attentively.
And then I find it kind of funny, in a very sad way, that there are people that expect the car to detect this sort of thing. Why do they think this? The car is not magic... No matter the detection system or manufacturer, any car on the road today can encounter situations where it will fail to stop or even react at all. This is just a really hard problem.
When combined with an attentive driver who is not relying on the assist in any way (i.e. behaving as if they do not exist), these systems (AEB, FCA, not including AP which is not relevant here) will be better than the alternative of just a driver with no aids. But at the moment, that is as far as it goes.
EDIT: if AP (or AEB) cannot react (when it is being misused) to a stationary fire truck, it seems unlikely it would react every time to a stationary pedestrian! Reacting to stationary objects I imagine is pretty tough since a lot of things (like the road if it is undulating) probably look stationary to a computer and neural net. Human brains are pretty impressive. If the pedestrian had been moving maybe it would have reacted...or not... Personally, I would not expect AP to stop me if I approached an unexpected traffic jam at freeway speed, if TACC was not following any one prior to the required stop. Have not had occasion to test it thankfully. It might work (I certainly get FCW alerts sometimes so it might help...but when that goes off (even on “early”) it is getting close to a crisis), but I think the chance of someone piling into me from behind subsequently would be higher than if I stopped myself, attentively.
Last edited:
Similar threads
