March of 9s: The Beginning of the End

I have had enough experience now with FSD 12 to say that in my opinion, I believe Ashok is correct:

https://twitter.com/x/status/1774673326583210476

I would also stress we are at the beginning of the end. That is, we will hopefully now start to see significant improvement with the "known FSD issues" and "corner case disengagements" with every major FSD release. "The end" (level 3 /4 autonomy within a wide Operational Design Domain (ODD)) is almost certainly year+ away. I'd like to document, just from one consistent anecdotal use case that can be repeated over time, where we are "starting from."

I will be driving a 90ish mile (2-3 hrs) "loop" under FSD to cover a range of driving scenarios. (Mileage is approximate):

For privacy reasons, there are about 15 miles in my loop that are not included in the link above that takes me to/from my actual home.

Here is a Link to inspect the route in detail

This route takes me from NJ Suburbs into and out of Manhattan (NYC) and includes approximately:

• 10 Miles Suburban driving
• 65 Miles "limited access highway" driving (This should be using the FSD 'highway stack' which is not the same as the FSD 12 stack)
  • includes interchanges
  • Includes Tolls
  • Includes Tunnel
• 8 miles of other "highway type driving", (will probably fall under the FSD 12 stack)
• 6 miles of dense city driving...including areas around Times Square, Rockefeller center, etc which will have dense vehicle and pedestrian traffic.

I will not be recording with a phone or anything like that. However I will try and save dashcam footage of anything notable.

I will report on:

1. Interventions (Accelerator presses, particularly if safety related)
2. Disengagements (comfort or safety related)
3. Overall impressions

As we know Version 12.3.x does not support...(but will need to in the future):

1. Smart Summon or "Banish"..so what I call the "first 100 yards and final 100 yards" is not available to test. (Drop offs / pick-ups).
2. "Reversing" while on FSD is not yet supported

Finally, there are what I would say 2 well documented "comfort / safety" issues with FSD 12.3.x that I have also experienced regularly first hand:

1. "Lane Selection Wobble"...for example, approaching an intersection where the single driving lane splits into multiple lanes (turning vs. straight)...FSD may act "indecisively"
2. Unprotected turn (stop sign) behavior. Notably: stops early....then creeps. If no cars detected it may creeps into intersection instead of "just going". Further, if it has creeped into the intersection, THEN detects a car approaching, it may still hesitate and require intervention (accelerator press) to get it going.

In addition to those two consistent issues, I expect to encounter some issues related to routing, and any number of other 'corner case' issues. All things that will ultimately need to be handled, but we expect to see dealt with as we progress though the "March of 9s"...toward the "end of the end".

Although I have driven FSD regularly over the past 3 weeks...I have yet to take it into NYC.

Vehicle: Refresh Model S (2023), Vision Only, HW4. First test will be using:
Firmware version: 11.1 (2024.3.15)
FSD Version: 12.3.4

So...there's the set-up. I expect later today to drive the first loop.

 

[Dewg]

People who self-diagnose with Asperger's Syndrome (known for being socially awkward) should not assume the task of naming products.

Imagine what he'll do with the social order on Mars. "Okay, everyone with an even number is an Eloi. Everyone with an odd number is a Morlock. Oooo, this is gonna be awesome!"

Of the 420 people in the original colony on Mars, 69% will be Morlocks.

 

[Joey D]

I've *sugar* on FSD my share of times and V12 is the first version where I've thought, "This can get there" as well.

It's really night and day from V11. Still has warts, but exponentially improved over V11. I want to wait until 12.4 to see how big the changes are with V12.

Exactly why I started this thread. Granted, my personal route and observations will be anecdotal...but I believe will be at least somewhat telling.

 

[Joey D]

https://twitter.com/x/status/1783925863295422732

More progress...

 

[EVNow]

The sheer convenience and cost benefits enabled by automated driving puts immense political pressure on the other side.

But - for that they have to be actually better than uber etc. Currently (Waymo etc) are not. They are just a curiosity.

 

[Joey D]

But - for that they have to be actually better than uber etc. Currently (Waymo etc) are not. They are just a curiosity.

At least has to be comparable in terms of comfort and efficiency....as long as it is significantly cheaper.

 

[lencap]

Thank you Joey D for your posts - very informative and helpful. FYI I have been using the FSD trial in NC, primarily in Raleigh, but also on longer interstate trips within the state.

As noted in Main Dashboard there have been a higher amount of DE listed for NC, and the small number of examples suggests that I may be disproportionally contributing to these reports. For that reason I'd like to comment on why I disengaged.

The great majority of my disengagements, like yours, aren't purely safety related, but rather how the car responds to more common driving situations. For example, several of my DE are related to driving too quickly in a given circumstance (accelerating from a stop, or making a sharp turn at higher speeds than I would typically drive). Another series of DE relate to construction issues - lanes not properly marked/detours/obstructions that are potentially dangerous, but not immediately so. The last major category include: debris in the road not recognized by FSD (a bumper, tree limb overhanging part of a lane). Of the roughly 20 DE I entered, maybe 2-3 were potential safety concerns, including one aggressively sharp turn that cut a corner and was hazardous. That was the only truly significant issue, and I took control immediately.

There were perhaps 3-5 other issues that were uncomfortable, but in terms of my judgement of a potential accident, or likely incident, they represented less than 10% of all DE, and some of them were likely not as dangerous as I initially thought.

I also noticed that the debris in the road issues were almost all on Interstate roads. It may be that those miles were governed by FSD 11 stacks, not the newer version 12 programming.

Overall, I thought the FSD experience would be similar to a 17 year old new driver, and expressed that to some other local Tesla owners after my first day or so of driving with FSD. In reality the more I used it the more comfortable with it I became. I began to realize when DE would likely occur, and shifted to manual driving during those times (traffic accidents with police directing traffic/unexpected road closures related to construction/accident).

I had intended to let my FSD trial expire and return to the standard autopilot. I may still do that (largely because I'm cheap), but on the other hand, as I wrote in another post, FSD is actually quite reasonably priced. I drive 10-12K miles a year, mostly local in town. Those drives average about 25 MPH, including traffic light stops, congestion, etc. The remainder of my driving is interstate, where I likely average about 50MPH, accounting for charging time/restroom breaks/etc.

So, if I drive 10K miles locally, and 2K miles interstate annually that equates to 320 hours of city driving time, and 80 hours of interstate driving - 400 hours a year. If I subscribe to FSD for an entire year at $99/month, the total cost is $1,200. For 400 hours of use that's $3/hour. Is is worth $3/hour to have the option of FSD, even if it isn't perfect? Right now I don't have enough experience to decide, but my initial reaction is that as a senior citizen who travels in all kinds of weather, and typically drives on unlit interstates at night, the cost of continuing to experiment with FSD isn't onerous.

I do note that my home charging expense of less than $0.035/mile for 90% of my driving, and perhaps $0.12/mile SC cost when on interstates, 10% of my driving. Over 12,000 annual driving miles that totals $380 for home charging annually, and an additional $270 for SC use, a total of ~$650. If I then add in the $1200 for FSD year round, my total "driving expense", consisting of "fuel" (electric bills), and FSD "chauffeured driving" totals less than $2,000 year, equating to ~$0.16/mile. That compares to driving my former ICE, which requires premium gasoline at $4.25/gallon and averages 25 miles gallon (combined city/highway) at a total gasoline equivalent price of just over $2,000, essentially equivalent to my combined EV charging AND FSD in my Tesla.

It's hard to ignore both the economic benefits of EV use (I haven't even covered ICE maintenance) along with the still improving benefits of FSD to a conventional ICE vehicle. Yes, FSD isn't perfect, and Yes again, it's not true FSD, but as an adjunct to everyday traffic situations in a fairly large city, so far I'm more inclined to applaud Tesla's new capability than criticize it. I know it's not perfect, but I also know from driving my 2018 Model 3 and my current 2023 Model Y that after tens of thousands of actual driving miles in the Tesla vehicles I tend to KNOW when I'll run across a situation where I need to take control, and that, combined with the more capable FSD version now available, make city driving far more relaxing that possible in other cars with or without their version of FSD.

I will also note that there was one time when my Model 3 (on autopilot not FSD at the time - 2020) slowed abruptly on the interstate when I was on autopilot at 70MPH. I didn't see that there was a car in the left lane (I was in the middle lane) that was accelerating very quickly and cut in front of me to take the upcoming interstate exit. Clearly the driver wasn't aware of the upcoming exit and made an unsafe cutoff in front of me, but it was late at night and I didn't see him at all! The Autopilot did, and the phantom braking wasn't phantom after all. How much is that worth? More than $3/hour.

I still haven't decided if I'll subscribe full time, but I certainly will experiment with monthly subscriptions for any trips/extended driving that I plan to do.

Thanks again for your reports and updates - they are very helpful.

EDIT NOTE: Corrected some math errors in initial post related to EV and ICE costs. The information above is now accurate.

 

[EVNow]

At least has to be comparable in terms of comfort and efficiency....as long as it is significantly cheaper.

Right - and safety too.

Safety can cut both ways - you might feel safer in a car without a driver if you are traveling alone at night (esp. for women). Or you may feel safer with a driver if the drive is through not a good area where someone can mess with the car.

 

[rlsd]

You can long press on the Satnav screen on the location of the "waypoint" you want to create, then click on the "+Pin" button to add the waypoint to your drive, altering your route. In many cases, it will add the waypoint in the correct order, but you may need to "edit route" and drag the waypoints into the correct order if you have multiple waypoints.

When using Waypoints with FSD, I have found that the car may want to "stop" at the waypoints...so you may need to accelerate through them or better yet, remove the waypoint from the navigation once you are getting close to it.

I wish FSD can give us an easy way to set the route that includes the streets that have the points of interest we want to go through without stopping. Also an option to exclude some streets.

Currently FSD routes me through some streets that have a lot of potholes during rush hours. FSD doesn't go through those streets when traffic are not bad.

Also an option to save the route with name. This will eliminate the time to setup for frequently used routes.

 

[Joey D]

Making logistical / progress in China as well:

Tesla Soars on Tentative Approval for Driving System in China

Elon Musk’s quick visit to China paid immediate dividends, with Tesla Inc. receiving in-principle approval from government officials to deploy its driver-assistance system in the world’s biggest auto market.

www.bloomberg.com

 

[Joey D]

My free trial has ended, so I will not be doing any more FSD drives until 12.4.x is available for me, at which point I will re-subscribe for a month.

After 30 days of heavy use of FSD 12.x, my basic conclusions:

2 Significant areas I am looking to be improved with the next major update or two:

1) "lane wobble / indecisiveness". Typically happens when approaching an intersection and the road "splits" into multiple turn lanes. The car will regularly exhibit "indecisive behavior" (start to change lanes...pause between lanes...and then move to a lane). While in this "indecisive" mode, it does not always "see" cars in front of you. I have had to disengage (brake or steer) several times for safety reasons as this is happening. This may be primarily a need to plan better, or just handle merges / lane changes in a more decisive manner. In any case...this is by far the largest reason for disengagements I have.

2) Unprotected turn indecisiveness. Dealing with this is usually not safety related, but comfort / efficiency. Approaching an unprotected turn when there is no cross traffic commonly takes too long to navigate through. It takes "too long" for the car to come to a point where it can "see", and then even though it is clear, takes too long to then commit to the turn. If there is no one behind you, it's tempting to just let the car do its thing (ad it will)...but clearly, if someone is behind you they would be getting impatient..so the proper thing is to intervene and accelerate through once you (the human) determines it is safe to go.

When there IS cross traffic, the car does efficiently pick a "good (safe) spot" to initiate and accelerate through the intersection / turn.

Other than those two items...I have come to EXPECT intervention free driving.

Elon has indicated that Number 1 is "fixed" with 12.4. (We will see.)

No indication yet on if / when number 2 is to be addressed. It may be that part of the issue is NHTSA insistence on certain behavior at stop signs that is preventing Tesla from handling these intersections in a more human like manner. Some believe that without side looking cameras that are more forward in the vehicle, it will be difficult to solve.

The 12.3.x versions appear to almost be completing the rollout, now that USS cars on AP3 are getting FSD 12.3.6. Hopefully, this means 12.4.x FSD rollouts will start within a couple weeks.

The rate of major software updates is as important as the quality of the updates themselves.