Version 17.4.14

[fasteddie7]

Interesting feedback. Was this mostly on divided highway?

My results are the practically the opposite. I feel like 17.4.14 is where I was 2-3 versions ago, as far as reacting to turns and staying between the lines, on undivided highways and country backroads. I know that is not the Tesla recommended use case, but I drive a route everyday that has become my baseline for testing AP. For example, there is set of 90 degree turns, smooth and level with clear lines, one 15mph the other 25mph. On 7.1, the car would not slow down at all into the turn, definitely would have run off the road if I did not take over. Early 8.0 releases I noticed it slowed down for one turn, but only in one direction. It wouldn't stay between the lines, but would at least stay on the road and recover. Later releases got progressively better, through 2.52.22, which would navigate both turns in both directions, albeit a bit sketchy, but the turn 7.1 almost handled was doable completely between the lines no intervention. I hoped with 17.x it would handle both turns completely.

But no. Now, the car is trying to kill me again. Barely stays on the road in that turn, and doesn't even attempt to slow down on the other one. I am hoping this is just a case of one step back, and we'll see an upcoming 2 steps forward!

On the freeway, things are great. But they have been for a while.

My ultrasonic arches do seem to extend further now, and the reaction time to move over when paralleling a wall is almost instant now, whereas before it would see it, but response was slow, if at all. I'll take that improvement!

As for the Dolby setting, mine turned off with the upgrade, but after turning it back on seemed to stick.

Haven't noticed much else yet, but looking forward to the future!

Same here. Autopilot is acting squirrelly again. The lines in my AP1 are jittery like AP2. More bouncing from line to line than I'm used to as well. It's as if tesla is updating the mobileye camera firmware to act like tesla vision cameras. Not sure if after a few drives it will get better, but today was unusual and AP acted more bizarre than I'm used to. In order to achieve parity with AP2 quicker, maybe they are reducing AP1 to meet AP2 halfway :/

 

[Hugh Mannity]

Same here. Autopilot is acting squirrelly again. The lines in my AP1 are jittery like AP2. More bouncing from line to line than I'm used to as well. It's as if tesla is updating the mobileye camera firmware to act like tesla vision cameras. Not sure if after a few drives it will get better, but today was unusual and AP acted more bizarre than I'm used to. In order to achieve parity with AP2 quicker, maybe they are reducing AP1 to meet AP2 halfway :/

Weird! Used the AP1 a bit today and it seems solid as usual. Did you try the two finger salute? You may need a reboot

 

[zambono]

I can confirm that the sticky mirror (not opening completely) has now been fully resolved. Despite the squealing I hear from my passenger side mirror during unfolding, the mirror is able to get to the right spot everytime! Cheers to Tesla. Perhaps was already solved in 17.2.3 a week or so ago.

Try to clean both open and closed where it could rub, and if that doesn't work try some lubricant / grease

 

[phaduman]

One other positive experience with this release:
The rush to get to set speed is now less aggressive. Say we are all going at 30mph in a freeway speed of 65. I have set my cruise to 69 (+4). The car in front moves to the right lane. The car ahead is say quite far up (because car right in front of me had slowed down to go to slower right lane). So now, there is quite a gap (way beyond the radar detection area, I have AP1) and in previous cases, the car would try to speed up rather quickly to try to go from 30 -> 69mph, and then detect after some time the car ahead and slow down aggressively.

This experience is way smoother now (both acceleration, and inevitable deceleration). My car is taking a much more sensible approach to increasing the speed and then slowing down more gently after finally seeing the car ahead. As if, the car knows I am in a commute time there is unlikely a chance that I would see no cars in front of me for long - so why speed up too fast? This is also safer as it helps reduce accidents from hopeful right lane cars trying to get into the left carpool lane I am on.

 

[LastGas]

My feeling is that the quicker you move to disengage Autosteer, the less force it takes.

 

[chillaban]

My feeling is that the disconnect force seems to be variable, almost as if it's based off confidence of the system (e.g. when it's less confident it seems easier to get it to disengage).

The other thing I've noticed: If Autosteer's opinion is that the steering wheel should stay still (e.g. following a straight road), it takes a good deal of force to disconnect it. However, if Autosteer's opinion is it should turn the wheel one way (let's say right) and your input is the opposite direction (turn the wheel left), there's almost no disconnect force. It seems like it weighs contrary input more than just deadweight on the steering wheel.

 

[phaduman]

The June 30 accident NHSTA investigation report talks about "microtorque" measurement technique to read driver engagement/takeover of steering wheel. I have also read somewhere that the autopilot steering torque is 4 lb/in (not sure about unit). I can see how Tesla would use all possible options to detect driver intervention. E.g. defeating autopilot in the opposite direction of steering movement is easier to measure than in the same direction.

Diff Q: what if and when those motors fail - how dangerous can they be, and how expensive would they be to replace? Perhaps no one knows yet - not even Tesla.

 

[chillaban]

The June 30 accident NHSTA investigation report talks about "microtorque" measurement technique to read driver engagement/takeover of steering wheel. I have also read somewhere that the autopilot steering torque is 4 lb/in (not sure about unit). I can see how Tesla would use all possible options to detect driver intervention. E.g. defeating autopilot in the opposite direction of steering movement is easier to measure than in the same direction.

Diff Q: what if and when those motors fail - how dangerous can they be, and how expensive would they be to replace? Perhaps no one knows yet - not even Tesla.

Which motors, the electric power steering motor?

A lot of modern cars use EPAS. Even the Toyota Corolla and the Ford Fusion. It's not a crazy expensive part though it is a $100k car and service will be priced like such.

As far as danger: not much. It's still just electrically assisted rack and pinion steering. If you lose the motor, the car will be very hard to steer as if you have no power steering. Of course Autopilot would promptly give you the flashing red hands disengagement.

 

[fasteddie7]

I have noticed since the update that my cabin seems more evenly heated. In November and December, it was around 5 to 15 degrees, and while driving I would always have cold legs and feet while the rest of the cabin was warm (I wear shorts a lot). The past few days after the update the temperature has been similar but I've noticed my cabin seems much more evenly heated... or it's my imagination.

 

[NullException]

Same here. Autopilot is acting squirrelly again.

Ditto.

For me, it's become more 'jittery'. I've noticed the car reacting more strongly to changes in traffic speeds, or things it thinks might be hazards, resulting in it making small adjustments to speed quite a lot on the highway where before it was very smooth. It's also abandoned lane changes on me several times and swerved back into the lane I'm trying to leave before starting to change lanes again.