basic autopilot: a bit disappointed

[DB288]

new MY

to me basic autopilot seems like adaptive cruise control that keeps you in the lane…

this is pretty common on new cars…esp luxury models …i have a 2017 accord that does adaptive cruise control and will resteer you if you cross out of the lane…not exactly as good but it is 6 years old and was much less expensive…

i am a bit disappointed…
———

i drive on some open country roads…

i use the autopilot…

my experience…if a car is slow or stopped in front of me …it will stop…

but if a light is red…it would blow through it..

i find it’s good for the highway but in other cases you have to keep turning it on and off…

 

[exxxviii]

This is exactly what Basic Autopilot is: Traffic Aware Cruise Control with Lane Centering. It is the same in in essence as every other ADAS out there: GM Super Cruise, Ford Blue Cruise, MB Driver Assistance, BMW Driving Assistance Professional, Toyota Safety Sense 3.0, VW Travel Assist, Rivian Highway Assist, Nissan ProPILOT Assist, Honda Sensing, Volvo Pilot Assist, Hyundai Highway Driving Assist, etc.

Tesla differentiates itself from all the others above with Enhanced Autopilot and FSD Capable. Those add semi-autonomous navigation to the core autopilot. FSD Capable has the ability to automatically stop at stop signs and stop lights.

 

[Febtober]

I mean yeah, that's exactly what they advertise it does. It shouldn't be any surprise. If you want it to do more, pay for enhanced autopilot or FSD.

 

[nate704]

Comparing Tesla AP with Honda or any other legacy auto's lane keeping is not even funny.

 

[arbitrage000]

Tesla’s real advantage over others comes with their FSD package to stop at lights and stop signs with no lead car and mostly with City streets beta which others don’t have anything like it.
Basic AP and even EAP don’t really do anything different than all the other systems. Although Teslas AP does function on all roads even without lane markings which other systems don’t. Some of the of the systems like Rivian are restricted to mapped highways.

If you want to get an idea of how Tesla regular AP stacks up against the others then OutofSpec YT has a series titled “hogback challenge” that gives an idea of how they compare.

 

[zoomer0056]

but if a light is red…it would blow through it..

Cruise Control in any car will not react to traffic lights.

 

[E90alex]

Basic Autopilot is actually pretty below average compared to many current drivers assistance systems from other automakers.

 

[karkrazy]

but if a light is red…it would blow through it..

You’re looking for FSD then - one-time $15k purchase or $199/month.

 

[zoomer0056]

i find it’s good for the highway but in other cases you have to keep turning it on and off…

This is what basic AP is supposed to do. It's for controlled access roads, highways. From the 2021+ Model S manual:

 

[exxxviii]

Basic Autopilot is actually pretty below average compared to many current drivers assistance systems from other automakers.

In what ways? I know a couple disadvantages, but those differences alone do not make AP "pretty below average." It really only lacks 1 feature missing from others. But that is offset by other Basic AP strengths.

 

[astrobill]

In what ways? I know a couple disadvantages, but those differences alone do not make AP "pretty below average." It really only lacks 1 feature missing from others. But that is offset by other Basic AP strengths.

LOL...in what ways does Tesla's "Autopilot" dissapoint? LOL

Let us count the ways:

- Since disabling/removing radar, it becomes totally non-functional in many rain and fog conditions that formerly were no problem. Much of the time, it pops up errors such as "XXXX camera blocked" and limits features, or more recently with all the fraudulent "vision only" Tesla Vision updates, the message "Autopilot speed limited due to limited front camera visibility"....forcing people to accept speeds as low as 35 mph on *Interstate* highways due to light rain...or sometimes just saying it's totally unavailable. This never happened back when my radar was fully used and before the ridiculous "Tesla Vision" project.

- It forces drivers to manually re-engage the lane keeping assistance after any lane change, which is totally ridiculous. Almost every other vehicle with lane-keeping assistance, including the all-electric Volvos and even the low-end Ford Fusion (LOL!) temporarily turn it off when the driver is signaling a lane change, then automatically turn it back on....it's easy to forget to turn it back on during long, straight drives in a Tesla and can lead to drift and accidents. Sure, one can pay Elon $6,000 or $12,000 (or whatever...he changes prices weekly) to get the "Navigate on Autopilot" upgrade, but then you have to deal with the vehicle deciding when to complete a lane change.

- Often ignores rows of cones when looking for lane lines and will barrel right into them. Tons of video of this and it's almost happened to me literally dozens of times over 2.5 years in my Model Y.

- Also...oh forget it, there are too many issues with it to even list. Vision-only was always a fraud, which is why Musk is putting radar back in this year. Removing it and claiming they could do "vision only" was always a supply chain based decision when they couldn't secure enough radar units during the height of the pandemic so he told his software team to somehow make due with just cameras. No other serious manufacturer is attempting to do "vision only" because of decades of testing, experience, and common sense. In fact, many are using BOTH radar and LIDAR in upcoming vehicles.

 

[E90alex]

In what ways? I know a couple disadvantages, but those differences alone do not make AP "pretty below average." It really only lacks 1 feature missing from others. But that is offset by other Basic AP strengths.

TACC is very rough and jerky. It races up to cars then slams on the brakes, then leaves like 5 car lengths before slowly speeding up again because it’s too timid. Low speed and stop and go traffic is very jerky with delayed response to the car ahead moving forward followed by too aggressive acceleration and then strong braking. I don’t have experience with radar Autopilot but my radar based adaptive cruise control on my Golf R was significantly smoother and better at following the car in front.

Phantom braking issues still.

Autosteer is too limiting in that it locks hard into the center of the lane and does not let the driver bias to one side or the other to avoid ruts/potholes or to give more space to oncoming traffic or big vehicles in the next lane. Trying to move it away from center cancels Autosteer. Changing lanes requires you to fight the wheel which cancels Autosteer and you have to manually re-engage again. The steering force required for hands on detection is very specific and it’s usually either not enough or too much and you cancel Autosteer.

And when not on Autopilot I find the lane keep assistance is barely passable. It doesn’t tell you you’re straying from the lane until you’re already out of it and barely tries to steer to nudge you back in the lane if at all. Most of the time it just crosses the line without any warning or correction or resistance.

Plus the annoyance of forcing on the (terrible) auto wipers when using TACC or Autopilot that go crazy berserk at the slightest sprinkle (or even with no precipitation because a shadow hit it wrong).

 

[exxxviii]

LOL...in what ways does Tesla's "Autopilot" dissapoint? LOL

TACC is very rough and jerky....

None of these things have happened with my Model Y; therefore your comments are just relevant to you and not reflective of the Basic Autopilot as a whole.

Not really, but it is important to recognize that forums have a very heavy negative bias, so we do not have a good sense of prevelance of things like Phantom Braking. (Oh, and other cars phantom brake too - it is not a Tesla thing.)

This is the first I have heard any one complain that TACC is jerky. In most objective reviews, the performance and capabilities of Basic AP are rated #1 or #2 among all the offerings. (Consumer reports ranked them #2.)

The one hard criticism above I agree with is the lane change annoyance with lane centering. Everyone else either automatically disengages and reengages or they just do the lane change. Tesla Basic AP forces you to disengage and reengage. And that does not even work that smoothly.

Oh, and Lane Keep Assist has nothing to do with ADAP (TACC + Lane Centering).

The grass is not as green as folks think. Tesla performs equal or better than almost everyone. GM is marketing the crap out of Super Cruise, to the point that people think it is comparable to FSD. Lots of people will be shocked to learn that it will only do lane centering on 10% of all US roads. BlueCruise is a little better, but it will only do hands-free on like 1% of roads.

 

[nvx1977]

new MY

to me basic autopilot seems like adaptive cruise control that keeps you in the lane…

this is pretty common on new cars…esp luxury models …i have a 2017 accord that does adaptive cruise control and will resteer you if you cross out of the lane…not exactly as good but it is 6 years old and was much less expensive…

i am a bit disappointed…
———

i drive on some open country roads…

i use the autopilot…

my experience…if a car is slow or stopped in front of me …it will stop…

but if a light is red…it would blow through it..

i find it’s good for the highway but in other cases you have to keep turning it on and off…

Objectively, the AP code is over 4 years old because Tesla decided they needed to rewrite the whole thing in order to make more progress. That rewrite is what is known as FSD beta today. Soon they will update the ancient AP code with the new code (you won't need to buy FSD to get it; it will just work on highways if you don't buy the extended offerings) and we should hopefully see AP behavior improve dramatically. Realistically, there may be some regressions to work through initially.

For those of us following along in the early days, AP was always only intended for use on highways. But Tesla had an anti-nanny culture and allowed you to enable it almost anywhere. It worked so poorly off highways that most of us just didn't bother. But in more recent years, AP started to work pretty decently on local roads. People who didn't RTFM assumed AP was designed for any road. But that's why AP doesn't stop for stop signs and stop lights; it was always only intended for highway use.

 

[SoCal Buzz]

This is the first I have heard any one complain that TACC is jerky.

I’ve had several generations of Tesla AP (currently a 2021+), and with the transition to Vision (my car has inactive radar now), it struggles in stop and go traffic. The main issue is it’s slow to recognize braking several cars ahead and stops too abruptly. Too the point I worry about cars behind me not paying enough attention. So in slow traffic I set follow distance at 6 to provide a buffer if needed, while 3 in faster conditions.

The latest AP does better job of accelerating from a stop when cars start to move, but overall I would say it is jerky in stop and go. At normal speeds and traffic flow it’s much smoother.

 

[E90alex]

None of these things have happened with my Model Y; therefore your comments are just relevant to you and not reflective of the Basic Autopilot as a whole.

Not really, but it is important to recognize that forums have a very heavy negative bias, so we do not have a good sense of prevelance of things like Phantom Braking. (Oh, and other cars phantom brake too - it is not a Tesla thing.)

This is the first I have heard any one complain that TACC is jerky. In most objective reviews, the performance and capabilities of Basic AP are rated #1 or #2 among all the offerings. (Consumer reports ranked them #2.)

The one hard criticism above I agree with is the lane change annoyance with lane centering. Everyone else either automatically disengages and reengages or they just do the lane change. Tesla Basic AP forces you to disengage and reengage. And that does not even work that smoothly.

Oh, and Lane Keep Assist has nothing to do with ADAP (TACC + Lane Centering).

The grass is not as green as folks think. Tesla performs equal or better than almost everyone. GM is marketing the crap out of Super Cruise, to the point that people think it is comparable to FSD. Lots of people will be shocked to learn that it will only do lane centering on 10% of all US roads. BlueCruise is a little better, but it will only do hands-free on like 1% of roads.

Autopilot is now ranked 7th by consumer reports.

 

[PVia]

People who didn't RTFM assumed AP was designed for any road.

Exactly...

The one hard criticism above I agree with is the lane change annoyance with lane centering. Everyone else either automatically disengages and reengages or they just do the lane change. Tesla Basic AP forces you to disengage and reengage. And that does not even work that smoothly.

EAP Auto Lane Change works perfectly...it's a joy to use.

Autopilot is now ranked 7th by consumer reports.

If you read their reports, you could see that they don't get it...

 

[exxxviii]

Autopilot is now ranked 7th by consumer reports.

You didn't read it... CR did a review with only one criteria for the actual performance of the system. The other four criteria are CR's subjective ranking of how much they liked the systems. Thing like driver engagement, ease of use, safe to use, and unresponsive driver. The problem with their rubric is that the five criteria are equally weighted. If you had read it, you would see that they ranked Mercedes-Benz as #1 in performance and Tesla in a group with Ford, BMW, and Toyota as the next best performers. By performance, they mean driving smoothly in the center of the lane in a variety of conditions. The rest are fluff by comparison.

 

[exxxviii]

If the rest are "fluff" by comparison and the others are subjective, it doesn't say much. I'm not impressed...

It was probably poorly worded. I meant to say that the other four subjective criteria are fluff by comparison to the one criteria about actual performance and capabilities. Those four reflect CR's particular bias toward their preferred user interaction experience, while the one performance criteria is an effort to somewhat objectively evaluate the system's ability to drive a car down the road. They should not have been equally weighted. Or even better, CR should have created a 2x2 matrix like a Gartner Magic Quadrant with something like "Ability to Perform" on one axis and something like "User Interaction" on the other axis that is a combination of the 4 subjective criteria.

The deeper takeaway from the CR review, when you look at their scoring and read their notes, is that there is huge parity among the top performers. In a nut, there is not much that separates M-B, Ford, BMW, Toyota, Tesla, GM, VW Group, Rivian, and Honda. Interestingly, they score Hyundai the lowest, and I am very familiar with theirs and it performs very closely to Tesla's Basic Autopilot. (Tesla crushes it on challenging roads and roads where you probably shouldn't use it, but they two are almost indistinguishable on long highway drives.)

 

[PVia]

Sorry, I deleted my original reply. I didn't want you to think I was dissing you in any way, when in actuality we basically agree.