Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

FSD Beta 10.69

This site may earn commission on affiliate links.
Is the safety score visible in the app? If so, it would tell you how many AP miles you have driven.
Oooo! It is. I think that disappeared last year after I first qualified. Though I'm failing to find AP miles. Anyway, I'll soon surpass 100 miles if I haven't already since the repair.

I see I got dinged hard for a forward collision alert yesterday that wasn't even my fault. I was in my lane, and a bike was in the bike lane. The cyclist was deliberately sweeping back and forth in this ongoing series of S curves, just for fun, I guess. He was staying wholly in the bike lane. But the car freaked, probably extrapolating one of his leftward sweeps as coming in front of the car. Oh, the injustice of a false conviction!

So if I in fact have to requalify, what is the current minimum safety score?
 
  • Like
Reactions: PACEMD
Oooo! It is. I think that disappeared last year after I first qualified. Though I'm failing to find AP miles. Anyway, I'll soon surpass 100 miles if I haven't already since the repair.

I see I got dinged hard for a forward collision alert yesterday that wasn't even my fault. I was in my lane, and a bike was in the bike lane. The cyclist was deliberately sweeping back and forth in this ongoing series of S curves, just for fun, I guess. He was staying wholly in the bike lane. But the car freaked, probably extrapolating one of his leftward sweeps as coming in front of the car. Oh, the injustice of a false conviction!

So if I in fact have to requalify, what is the current minimum safety score?
Got in in May with a 92. Rumor has it that it'll open to people with 80's in the near future, but who knows?
 
Just got FBD beta today.

I have to say, the people who say it’s cool and the people who say it’s dangerous are both 100% correct. Very cool proof of concept. It also blew a stop sign, drove into the middle of a major roadway, and just stopped there, within the first 15 minutes of my first drive, lol.

If I remember correctly, Elon said it would be 200-300% safer than a human driver by next year. IMO, there’s a 0% chance of that. But, I can see it being the future of driving sometime in the next 5-10.

My $0.02.
 
People complain about how FSD handles unmarked roads with driving down the center, but it is like it was designed to work with "advisory bike lanes" where both directions of traffic drive down the center of the road while bikes get to use the lanes on either side:

1664593177102.png


Some of the stupidest road design I have ever seen.
 
People complain about how FSD handles unmarked roads with driving down the center, but it is like it was designed to work with "advisory bike lanes" where both directions of traffic drive down the center of the road while bikes get to use the lanes on either side:

Some of the stupidest road design I have ever seen.
I thought this too when I watched a video about it but........it is about the SAME as driving an unmarked road except MUCH safer. On unmarked roads you have VRU's on both sides roads, often occluded and must navigate. On a road with advisory bike lanes you aren't required to drive in the middle (stripped and NOT solid) and can move to the side but MUST yield to VRU's. The lines are just there as a visual cue that guides you away from the MUCH more dangers sides of the road which are highly likely to have VRUs. So while at first it looks like a stupid or counterintuitive idea, in actuality all unmarked roads would be much safer if they were painted this way. It is totally impractical to just widen all residual type roads to make them safe for VRUs and have 2 lanes of detected car travel.

To me the problem with unmarked roads is not that Beta drives near the center (which is statistically safer) but when another car is opposing it is often late in giving the other car a visual cue that it will start moving over. It tends to wait too late (like playing chicken with the other car) and sharp/jerk veer to the right instead of gradually moving to the right the way a human does.
 
  • Like
Reactions: sleepydoc
So,
The last three days, I’ve had “Whole Mars Catalog” drives. No interventions or disengagements. Even did a “Chuck style” unprotected left turn, used the center turn lane, did great. A couple odd things that need some work, but all in all, pretty decent.
To follow up on this, this week has been the complete opposite. Absolutely awful drives. Seeing a lot of the same issues people are reporting.

The main issue I’ve seen is the car pulling out for a turn like it should then stalling mid turn, in front of oncoming traffic. Left or right turn, doesn’t matter. I intervene with the go pedal. Doesn’t do it always, but when it does, it’s pretty awful behavior.

I’ve found this on unmarked roads: the width of the road plays a part on how the car interacts with other cars and VRU’s.

Example: I live on a rural unmarked road, which I’d say is “narrow”. My car comes to a damn near complete stop for on coming traffic or if someone is walking by.

A few miles away, there’s another unmarked road, but much wider. No issues at all, does great.

It stops/swerves for deer as well. Renders them as what looks to be Great Danes on the display. Had this happen last night, my wife was surprised at how well it reacted to the deer in the middle of our road. Speed was about 37mph.
 
Y'know, I'm beginning to wonder. The general idea has been that we get the firmware download, it has its algorithms, neural and otherwise, and we drive the vehicles around the landscape, seeing how the algorithms do, reporting back to Tesla with the video icon as we go.

There's also those of us whose data gets uploaded to Tesla on demand; so, for example, they can compare how we drive vs. how the car would drive. And any such inspection could easily be done sans human hands or eyeballs, all the better to figure out how the algorithms are working.

However, I'm going to add a snivvy: It might be possible that parameters of the system software are being modified by Tesla over time. So, one day, one wakes up and the parameters are that way; the next day, one wakes up, and the parameters are this way. Why do this? To get an idea how the parameter variation affects the car's autonomous driving performance, of course. Done across a large fleet, different areas, different weather, different traffic loads, this would possibly give Tesla even more data on how to Make It All Work.

And might explain why some days everything seems to work beautifully and other days, not so much.

If Tesla is doing this (and that's a big if), well, I guess I don't mind. We're out here doing our testing bit, that's what we've volunteered for. I suppose that the car and driver could get into Big Trouble this way; but that's true of driving around in a true beta software load in any case.

Comments?
 
Y'know, I'm beginning to wonder. The general idea has been that we get the firmware download, it has its algorithms, neural and otherwise, and we drive the vehicles around the landscape, seeing how the algorithms do, reporting back to Tesla with the video icon as we go.

There's also those of us whose data gets uploaded to Tesla on demand; so, for example, they can compare how we drive vs. how the car would drive. And any such inspection could easily be done sans human hands or eyeballs, all the better to figure out how the algorithms are working.

However, I'm going to add a snivvy: It might be possible that parameters of the system software are being modified by Tesla over time. So, one day, one wakes up and the parameters are that way; the next day, one wakes up, and the parameters are this way. Why do this? To get an idea how the parameter variation affects the car's autonomous driving performance, of course. Done across a large fleet, different areas, different weather, different traffic loads, this would possibly give Tesla even more data on how to Make It All Work.

And might explain why some days everything seems to work beautifully and other days, not so much.

If Tesla is doing this (and that's a big if), well, I guess I don't mind. We're out here doing our testing bit, that's what we've volunteered for. I suppose that the car and driver could get into Big Trouble this way; but that's true of driving around in a true beta software load in any case.

Comments?
I was just going to comment on @LowlyOilBurner ’s post about how inconsistent 69.2.2 seems to be. Your theory would help explain that inconsistency.

I’ve had several turns like s/he describes where it would be fine except it suffers from a failure to commit, forcing me to stomp the accelerator. Other times it does beautifully. 🤷‍♂️

Anyone who’s driven FSDb should know full well that disclaimer ‘it can do the worst possible thing and the worst possible time’ is completely correct.
 
  • Like
Reactions: ironwill96
However, I'm going to add a snivvy: It might be possible that parameters of the system software are being modified by Tesla over time. So, one day, one wakes up and the parameters are that way; the next day, one wakes up, and the parameters are this way.

Comments?


Nothing except a full firmware upgrade can survive a reboot- so that'd be a pretty fragile system.

The only thing that can consistently improve the behavior of the system other than a full firmware upgrade is a map update (which can be done entirely silently/in the background, as compared to the firmware update)
 
Nothing except a full firmware upgrade can survive a reboot- so that'd be a pretty fragile system.
True but it's certainly not inconceivable that Tesla has various parameters that can be tweaked/pushed without notification - i.e. change this value from 14.6% to 14.9%, affecting positively component A with some potential negative impact to other components. Easily pushed over any network, very small in bandwidth. Those could hypothetically survive a boot. Sure you likely don't introduce any new features this way but modification of parameters designed to be adjustable certainly could be. Perhaps @verygreen knows better but it actually sounds like a smart approach to me.
 
True but it's certainly not inconceivable that Tesla has various parameters that can be tweaked/pushed without notification - i.e. change this value from 14.6% to 14.9%, affecting positively component A with some potential negative impact to other components. Easily pushed over any network, very small in bandwidth. Those could hypothetically survive a boot.

No, they can't survive a reboot

Sure you likely don't introduce any new features this way but modification of parameters designed to be adjustable certainly could be. Perhaps @verygreen knows better but it actually sounds like a smart approach to me.

He's the one who told us a while back this can't be done (and in some detail why)--

 
I’m one of the most recent additions into FSDb.

One thing that I’ve noticed that is divergent from all of my previous AP experience and something that I so far really dislike:

If I’m going say 45 in FSDb and I see something ahead that I want to slow down for, previously I would scroll the speed down with the right scroll wheel, say to 25. In my previous experience with AP the car would gradually slow down. With FSDb the car seems to completely ignore the right scroll wheel input IF speed is reduced from current speed.

Anyone else experience this?