Is the current NoA helpful or harmful?

[M109Rider]

These reports by these media companies, are designed to generate views just like any other news outlet, by reporting in a way that solicits response.
They aren’t based on valid reporting.
I thought consumers reports was different, but they seem to be going the way of every other news media now. Very sad.

 

[Rogue one]

I drive 200-300 miles a day in a work truck the number of human drivers I see performing unsafe maneuvers is staggering. People pass on the right all the time sometimes even passing in exit or entrance ramps. The people who tested this for CR must live in a utopia where everyone follows the law to the letter because where I live and drive often (Milwaukee area) people drive significantly worse than my navigate on autopilot ever has. Yes sometimes it makes mistakes that’s is why you need to monitor it I find monitoring it significantly easier than driving. The point is people make significantly more mistakes or dangerous road conditions than NOA in my opinion and observation.

 

[RxShah]

Respectfully, I haven't been able to use NOA successfully in the DC Area on the major highways. I think it probably does a great job in some areas, but it is harmful and puts me in harms way in this area. I appreciate everything regular AutoPilot can do and does, but NOA is not ready for primetime (everywhere at least).

I use every time I drive 230 miles between DC and New York, I know I have to take over at toll booths and construction zones. But I find it more relaxing that Autopilot. It drives and takes interchanges successfully.

 

[jananglin]

I agree with CR’s account of NoA but they should have made it clear that regular Autopilot (lane keeping, TACC, LDW) is best in class and is what is enabled by default.

If I were a casual observer reading the article I’d be under the impression that buying one of these cars means it will just start driving itself into dangerous situations like some sort of suicidal robot.

NoA for me has gone from completely terrifying (November) to scary-and-irritating (January) to a system that’s useful in extremely specific circumstances (today) as long as you’re ready to regain control immediately. You can see where NoA is headed and in the right situation it’s magic, but I think they gave the option for it being on automatically at every Autopilot engagement too soon. It needed more time in the oven.

 

[notrhj]

NJ the law is Keep right except to pass.
If your car is constantly passing on the right maybe you’re the problem ?

 

[mswlogo]

I think it's the way we use NOA wrong. I found it's way too aggressive on higher settings and need a lot of extra attention because of constant lane changing.
However, now I set
- the NOA setting to mild,
- following distance to 6 cars,
- and set speed to be just at the speed limit or slightly slower than the car in front of me.

Then it works really well, only suggesting changing lanes when there's a slow truck in the front, then it will change lane back after passing. I just drove 300 miles yesterday and feels it helps quite a bit.

I had mine set to mild initially with 3 co-workers. They were laughing their heads off “nobody drives like this”. They were more impressed when I set it to mad max.

I find it does not do well if I want to chill. If a car in front drops 1mph from my set speed. I don’t instantly want want to pass them. I wait a bit. But when I do decide to pass I do it swiftly. Currently you can’t have both.

It kind of needs hysteresis or something. It’s tricky. Because you decide if this is a good person to cruise behind or not. Even if they wander a bit. But if they do something odd or slow just a wee bit to much, we are out of there.

Maybe it needs a pass now button.

 

[t-rizzle]

I think it's the way we use NOA wrong. I found it's way too aggressive on higher settings and need a lot of extra attention because of constant lane changing.
However, now I set
- the NOA setting to mild,
- following distance to 6 cars,
- and set speed to be just at the speed limit or slightly slower than the car in front of me.

Then it works really well, only suggesting changing lanes when there's a slow truck in the front, then it will change lane back after passing. I just drove 300 miles yesterday and feels it helps quite a bit.

When you say set your speed you mean manually set it at the posted speed limit that the car would recognize correct? Wanted to make sure you were not referencing the speed limit settings in the AP settings, maybe Driving settings, as I still don't really understand that yet.

 

[jananglin]

I believe the CR report is accurate. My wife and I own a 2017 Model X. Goofy lane change suggestions, "phantom braking" and entering an exit too fast are my biggest concerns with NoA. I only initiate lane change suggestions. While traveling down a country road using NoA the car would rapidly slow for mail boxes near the road's right shoulder so my "alert antennas" pop up whenever anything approaches and encroaches from the right. NoA may not be ready for prime time but regular Autopilot, now, for me is a regularly used feature. I use a combination of the two features about 80% of the time I'm driving. What I've pleasantly discovered is that the stress and exhaustion I generally get from driving long distances, sometimes in poor weather, has been drastically reduced. Now comfortable driving days are about 10-12 hours long and there have been a few that were 16 hrs +, I don't recommend those. To sum it up, I think all the driver assisted features require supervision and infrequently the appropriate intervention but the car: drives better than I drive, has increased my travel range about 100%, keeps me comfortably alert while traveling in situations of reduced visibility-heavy rain accompanied with road spray-I don't even want to revisit the anti-slip independent wheel traction control that kept me from falling about 2,000 ft. into Pendleton OR...maybe it doesn't enter all exits too quickly. The car keeps getting better.

 

[ljbad4life]

There is definitely some truth to the CR report. NoA is improving and can get "squirrelly" in urban areas. For now it's for long drives on highways that are up to national standards, not ones grandfathered into the Interstate system.

 

[Baltimoron]

It's all about context. Does it require intervention now and then? Yes, Absolutely.

Does it make my commute in traffic safer because it is more attentive than me? Yes, Absolutely.

I'm in this to watch the improvement over time, and experience this once-in-a-lifetime transition to automated driving. As an engineer (unrelated to auto or self driving), I find this whole thing fascinating.

I couldn't agree more. Well stated.

 

[Runebane]

I have not read all the replies on this thread (I know I'm a horrible person). That being said, I agree with CR to a point, that NoA does require a lot of user intervention. But what struck me as odd is they said NoA cut someone off... NoA is such a granny that doesn't seem possible and I personally would need to see a video of it happening because I do not believe that NoA can cut anyone off with it's current programming. If it wants to change lanes and someone a half mile away speeds up, it waits for them (ok, slight exaggeration) but seriously that's one of the reasons I have to intervene so much is because it's TOO hesitant to change lanes.

The other thing, but this applies to just AP, if there's construction and the lines get covered up, it has no clue what to do. A few updates a go it actually would see the yellow line and swerve to follow it, it doesn't do that as much through the same construction area now, but occasionally it will. That can be dangerous if you're not paying attention or don't have your hand on the wheel when it suddenly jerks to "center" itself. But it is a work in progress and labeled as BETA not turned on be default. These are things the article should have explained as well.

TL;DR - Yeah it requires intervention, but the article writer also apparently requires intervention as well.

 

[SSonnentag]

Here's my 2 cents:

NoA is safe, but it requires so much user intervention to prevent unnecessary slow-downs, speed-ups and delays while waiting to detect hands-on-the-wheel that I choose not to use it. Again, safety isn't an issue, but it certainly doesn't drive more smoothly or fluidly than a human, or at least this human. I think it doesn't "look" far enough ahead to make intelligent choices on when to change lanes, so it makes changes when it shouldn't and doesn't when it should, but it's not unsafe about it, just dumb. For instance, NoA will move left to pass a string of 5 semi trucks when it is only 1/4 mile from the exit it needs to take. I can see there is no space between the trucks and will sit out the 15 seconds it takes to follow the trucks to the exit point. NoA will just attempt to pass and then get stuck in the left lane as the exit goes by.

 

[chickensworth]

I think it's the way we use NOA wrong. I found it's way too aggressive on higher settings and need a lot of extra attention because of constant lane changing.
However, now I set
- the NOA setting to mild,
- following distance to 6 cars,
- and set speed to be just at the speed limit or slightly slower than the car in front of me.

Then it works really well, only suggesting changing lanes when there's a slow truck in the front, then it will change lane back after passing. I just drove 300 miles yesterday and feels it helps quite a bit.

that sounds absolutely painful... 6 follow distance and mild NOA settings... maybe if there are no other cars on the road.

i find mad max to be way too conservative... constantly being overtaken and losing opportunities to change lanes. in any medium or heavy traffic, NOA is more or less useless to rely on for changing lanes at appropriate times. you end up being the guy driving with his blinker on swerving back and forth for miles on end.

 

[Yangy]

That's the thing, it's a little bit counter intuitive.
If I drive myself, I might be more like mad Max mode. However, I don't want to "monitor" another mad max driver (the car) for a long time, it requires a lot of attention. I'd rather monitor a granny driver and take over only when it needs to be aggressive. With this pattern I can relax 90% of the time.

NOA right now basically can only change lanes, passing, or go off ramp, when there's either low or no traffic, or the other driver on the road yield to you when your turning signals are on. I found lots of other people don't yield. NOA is kind of shortsighted and slow to take action, it doesn't matter if you set settings to max or mild, the maneuver is the same. That's in the fine print of the use manual. Mild only makes it wait a bit longer before suggesting lane change, so that avoids the problem of stupid moving back and forth thing. Same for speed, I found if the car before you are within -5 miles than your cruise target, with mild settings it won't suggest passing it that often. So you just need to adjust the cruising speed to match traffic.

In my opinion the real missing piece of NOA is it can't set own cruise speed to follow traffic yet, some times you need +/- 10 miles to follow the traffic.

 

[GolanB]

In my opinion the real missing piece of NOA is it can't set own cruise speed to follow traffic yet, some times you need +/- 10 miles to follow the traffic.

That is a solid suggestion, and one that I would welcome.

In real life:

- I tend not to want to be within a cluster of vehicles and prefer to be away from other cars if possible.
- Its important that I not travel in other drivers blind spots.
- I tend to follow a lead car that is going slightly above the speed limit. I am OK with being a lead car for others.
- I tend to change lanes only when I can't maintain a relative constant speed.
- When traffic ensues, I'll re-adjust my expectation. If traffic gets bad, I don't bother passing other cars. My experience is that I won't gain much.
- I try to take traffic conditions on the roads into consideration when planning a route.

Some nice to have features for distance travelers:

- When traveling in excess of the speed-limit, automatically adjust based on reports of accidents, hazards, and police activity.
- Add advanced route planning natively. Traditionally I've used WAZE when computing future trips because it takes into account traffic and total travel time. This might influence where, when, and for how long we supercharge.
- Adjust driving to avoid remaining in other drivers blind spots.

 

[chickensworth]

That's the thing, it's a little bit counter intuitive.
If I drive myself, I might be more like mad Max mode. However, I don't want to "monitor" another mad max driver (the car) for a long time, it requires a lot of attention. I'd rather monitor a granny driver and take over only when it needs to be aggressive. With this pattern I can relax 90% of the time.

NOA right now basically can only change lanes, passing, or go off ramp, when there's either low or no traffic, or the other driver on the road yield to you when your turning signals are on. I found lots of other people don't yield. NOA is kind of shortsighted and slow to take action, it doesn't matter if you set settings to max or mild, the maneuver is the same. That's in the fine print of the use manual. Mild only makes it wait a bit longer before suggesting lane change, so that avoids the problem of stupid moving back and forth thing. Same for speed, I found if the car before you are within -5 miles than your cruise target, with mild settings it won't suggest passing it that often. So you just need to adjust the cruising speed to match traffic.

In my opinion the real missing piece of NOA is it can't set own cruise speed to follow traffic yet, some times you need +/- 10 miles to follow the traffic.

interesting take... i can see how it might eliminate some of the frustration and false lane change signals... maybe at the expense of making all the exits, merges etc... but worth a shot.

 

[SparkyB]

My experience is that NoA is not good in heavy traffic (Rush hour) in Dallas Texas. With moderate to light traffic it work fine. I have a 10 mile highway run I do 2-4 times a day. The three problems I noticed are:
1. Large trucks encroaching on my lane cause the T3 to brake hard even if the front of the truck was only a few feet away. As you know some truckers don't feel the need to stay in a lane with an 80,000 pound behemoth behind them. It is so fast that I wonder if I'll get hit from behind some day.
2. When trafic merges in from the right the T3 tends to drift toward the merging traffic because it thinks it has a wide lane and tries to center. Not a good idea.
3. My T3 keeps trying to get the the right lane... even of that lane is about to disappear. Generally good. I just cancel the Recommended Lane Change but It should probably try to stay in the current lane unless the lane change is requested or approved. Or, have a switch that only allows the car to pass on the left. This is the source of the passing on the right that some people have noted. If the left or middle lane is slow then when the T3 finds an open lane it tried to accelerate to the set speed. Don't know what the do in Germany it is illegal to pass on the right.

That said, I have taken an couple of road trips on the highways in Texas. For these trips the NoA has worked flawlessly 99.9% of the time. It is a great help on road fatigue. It always bears watching but in easy situations it works very well in my opinion.

 

[jandjmh]

I agree with CR’s account of NoA but they should have made it clear that regular Autopilot (lane keeping, TACC, LDW) is best in class and is what is enabled by default.

If I were a casual observer reading the article I’d be under the impression that buying one of these cars means it will just start driving itself into dangerous situations like some sort of suicidal robot.

NoA for me has gone from completely terrifying (November) to scary-and-irritating (January) to a system that’s useful in extremely specific circumstances (today) as long as you’re ready to regain control immediately. You can see where NoA is headed and in the right situation it’s magic, but I think they gave the option for it being on automatically at every Autopilot engagement too soon. It needed more time in the oven.

 

[jandjmh]

Yep - the current state of navigate on autopilot is very much like driving with a teen ager who got their license a few months ago. But before the recent updates it was like driving with a kid who got their learner's permit yesterday. The rate of improvement is tolerable. I do find Navigate on Autopilot quite convenient and relaxing under conditions of open roads and good conditions. I often let it practice under more difficult circumstances, but I am well aware I may need to take over. And if I understand how the learning system works, every time do intervene (which is pretty often) it helps train the system for everyone.

 

[Spalding]

What makes you think the CR guys don't? They have had a permanent Model S and Model 3 for a couple of years now and have continuously reported their findings, including e.g. uncovering a problem with the Model 3 braking system. They have also reviewed NoA when it first came out.

Not that CR doesn't know the product (although that's an assumption) - rather, their article was written in a way that makes the problem seem, to the uninitiated or to click-bait seeking web sites, as a condemnation of the entire Tesla self-driving process as dangerous flaw worthy of Ralph Nader. Yet the CR author even seems to disagree with the way the article is being portrayed in countless links:
Consumer Reports Not Thrilled By Misleading Navigate On Autopilot Headlines

No, I don't think Nav on AP is a fully baked product yet. I agree with some of CR's conclusions. But I don't agree with the impression created in the article that using Nav on AP invites firey death. The people on this web site know the product well and are able to put the concerns voiced in the CR article into scale and context. But the countless number of non-owners who read (or mis-read) the CR article, and the countless breathless descriptions of it in other web sites, won't have that knowledge.