Autopilot speeding ticket

[Dameon]

I have an AP1 loaner right now. My AP2 is in the shop. Definitely slows down at the signs. Right at the signs.

Just to clarify before I go through the effort of videoing this scenario. I have videoed other scenarios to support my debate. Most recently was the "bug" that AP1 drivers claimed if AP was engaged and slowing down to accommodate a car sitting at a red light, and another car switched into your lane between you and the stopped car, AP wouldn't brake properly. It does.

Clarification of test to be performed: I intend to engage AP1 on a roadway with a 40mph posted limit. I will engage AP1 at 50mph well before the limit reduces to 40mph. You state that the car will change my 50mph set point to 40mph at the posted limit (assuming it is able to read the sign). I would be willing to accept that it would make this change before/at/after the sign to eliminate the proximity of the sign being a factor. My prediction (based on near daily experience) is that AP1 will stay engaged at the speed I set, but will show that the posted limit is 40mph. I also predict that it will enlarge the 40mph posted limit icon to indicate I am exceeding the posted limit. Is this an acceptable outline?

By the way, this is not directly specifically at brkaus, I'm just tagging on to his response since it was the latest one in the thread.

 

[brkaus]

@Dameon - my AP2 car does slow right at the sign, assuming it is a single lane road that has AP speed restricted. If it is a double lane, restricted access, or otherwise I am allowed to drive more than 5mph over the limit than nothing happens.

I just turned in my AP1 loaner, so I cannot double check that car.

I don't need video evidence of anything. But a detailed description of all the nuances and what happens would be useful for all.

 

[JohnnyG]

Just to clarify before I go through the effort of videoing this scenario. I have videoed other scenarios to support my debate. Most recently was the "bug" that AP1 drivers claimed if AP was engaged and slowing down to accommodate a car sitting at a red light, and another car switched into your lane between you and the stopped car, AP wouldn't brake properly. It does.

First, on this point.... I'm truly glad you've never had to experience (or noticed) a close call. However, others have, and it can be terrifying.
My experience is that 95% of time it does an acceptable job. However, just last night, driving home from picking up my daughter from school, a car chose to change lanes in front of me; literally inches from my bumper. My car did nothing. How do I know, you ask? Because I have become complacent and wasn't paying close attention. He was very slowly, but surely coming over, and was 3/8 of the way over before I noticed. I didn't react at first; other than to regain 100% situational awareness. At the 1/2 way mark, I disengaged. Had he hit his brakes, or even just came off the throttle, we would have collided. Lesson learned for me... there's still a 5% chance that it won't react properly.

Clarification of test to be performed: I intend to engage AP1 on a roadway with a 40mph posted limit. I will engage AP1 at 50mph well before the limit reduces to 40mph. You state that the car will change my 50mph set point to 40mph at the posted limit (assuming it is able to read the sign). I would be willing to accept that it would make this change before/at/after the sign to eliminate the proximity of the sign being a factor. My prediction (based on near daily experience) is that AP1 will stay engaged at the speed I set, but will show that the posted limit is 40mph. I also predict that it will enlarge the 40mph posted limit icon to indicate I am exceeding the posted limit. Is this an acceptable outline?...

On this point.... I too can make a video, along with hundreds of others, demonstrating that on non-limited access roads, when AP is engaged (not just TACC), it will slow down to within +5MPH of the posted speed, assuming it reads the sign. My prediction is also based on near daily experience.

 

[Dameon]

On this point.... I too can make a video, along with hundreds of others, demonstrating that on non-limited access roads, when AP is engaged (not just TACC), it will slow down to within +5MPH of the posted speed, assuming it reads the sign. My prediction is also based on near daily experience.

I'll go through the effort of creating the video then.

 

[fasteddie7]

I'll go through the effort of creating the video then.

No need the video I posterd here of AP1 does just that. If you look at my other videos comparing AP1 and AP2 you can see both slow down to whatever speed the sign says plus whatever the offset it set to at the moment the nose of the car passes the sign Onlyinterstates and a few other roads does it not do this.

 

[bob_p]

However, just last night, driving home from picking up my daughter from school, a car chose to change lanes in front of me; literally inches from my bumper. My car did nothing.

We've seen this in the past week in our X with AP2.5 while running under AutoSteer.

When running at speed limit + 5MPH, there will often be other cars approaching from the rear wanting to go much higher than the speed limit. And one of those drivers decided to make a point of their frustration of how "slow" we were driving, and passed us with only inches to spare on the right side, cutting just barely in front of the car.

No warning...

This is going to be a bigger problem when Tesla has the "Navigate on Autopilot" feature with automatic lane change enabled. AP will keep the vehicle at what the software believes is the speed limit (plus some) - and there will always be a few aggressive drivers who want to go much faster - and drive closer than it is safe when passing.

Not only should there be an audible & visible warning when vehicles are too close - but the AP software may need to respond by automatically shifting slightly away from the vehicle inside the current lane, and either slowing or speeding up to ensure a safe gap between vehicles.

 

[Dameon]

Only interstates and a few other roads does it not do this.

There is a caveat to the statements above that no one clarified: "some" roads. I wonder what dictates this. In a smaller, possibly more rural, town like Lafayette, LA, verses a metropolitan like Los Angeles, CA, might not have the geofencing or whatever metrics to dictate the behavior. I may never see it on roads in my area. Indeed I was unable to capture the behavior stated by others on several roads this morning in my area, none of which were Interstates, all of which were 40mph to 50mph posted limits.

Also, there seems to be some missing of my point. Expected functionality verses failure of known functionality. We know the car is designed to detect objects entering the car's proximity and act accordingly (like a car cutting you off). It has been stated above that it the failure to do so may be a small percentage (5%?), but the point is the same. My point was that my observations and testing have repeatedly shown that my car, in my area, doesn't exhibit the behavior. Empirical evidence would dictate that the functionality is not there.

Since other areas do seem to exhibit the behavior, it appears to be a regional function. My vehicle cannot match the required parameters to exhibit the behavior. Determining the exact data points not being met in my situation has no bearing on others experiencing the behavior and I don't care enough to continue the effort of explaining it or attempting to determine those data points, so I'm done with this topic. It was nice to see how quickly the forum community jumped on me concerning my conclusion without understanding the possible exceptions that they may not be faced with.

 

[JohnnyG]

...It was nice to see how quickly the forum community jumped on me concerning my conclusion without understanding the possible exceptions that they may not be faced with.

Excuse me?? We "jumped" on you? You didn't start off you communication with "Hey guys, I experience something different than you guys! Let's discuss and see if we can figure out why! "

Instead, you insinuated that we were flat-out wrong, and you blatantly stated this....

...Most recently was the "bug" that AP1 drivers claimed if AP was engaged and slowing down to accommodate a car sitting at a red light, and another car switched into your lane between you and the stopped car, AP wouldn't brake properly. It does. ...

We didn't "claim" anything, we stated it; because it actually happened. And in our cases, no, it didn't brake properly. Again, you could have said something like "Weird! I haven't seen that before; tell me more, so I can expand my knowledge of these things!" Instead, you go on a diatribe as to how you are going to prove us wrong with your videography.

No one is saying that the system only does one thing and is consistent at it. On the contrary! Exactly what we're trying to point out is; it has a set of rules, however some input or logic is causing it to not be consistent, and that is the concern.

Not reacting to a car cutting you off; my only thought on this one is with AP1 (and AP2.x prior to fully utilizing the wide angle lens), the car is creeping in from a blind spot. Think of the IC display when a car gets just lined-up with your front bumper; it disappears.

Not dropping speed when in AS and reading a new lower speed sign; this one may be easier as there's only one criteria (that we're aware of)... The AP system groups roads into two classifications: limited access, and other. If the classification is wrong, it won't act as expected.
I've seen this both ways. There's a long, high speed road which has a median (direction of traffic is separated by a grassy area), but still has cross streets. This road appears to AP as limited access, even though it is not. This (and variations of this) are more so the case when it gets it wrong. There's a lesser case, where a road is limited access, but a minor state route, which the AP system incorrectly classifies as not limited access.

Anyway, just some thoughts and observations... I'm happy to hear what others have experienced.