Search results

I used FSD in BC; V11 worked - it also appears to understand the blinking green traffic light, which I don’t.

Anti-lock brakes are a requirement because they do work. Lidar/radar efficiency is not substantiated by anything other than numerous complaints about their removal.

I don't think it's easy to avoid any death linked to inattentive use. What worries me is that the press spins this as a failure of technology. The voters may press the regulators to change their minds and to make progress unreasonably hard.

How about this one? Though the driver admitted he was inattentive while using Autopilot, the preventable death still sounds unnerving. https://www.heraldnet.com/news/tesla-driver-on-autopilot-caused-fatal-highway-522-crash-police-say/

It's becoming more and more humanlike

With FSD, this happens to me when I don't wash my car after dusty roads. It is usually not able to change lanes if cameras are covered with dirt just enough not to trigger 'camera occluded' message.

It took Comma a few years to solve the laneless driving, I guess Tesla would take about a decade to do so.

I thought this was a recent accident, and the possible reason immediately came to mind. The latest FSD (11.4.9) includes a new "feature" of reading speed limits through OCR from nearby bushes. When driving at night in our neighborhood (all 25mph roads), my car constantly reads speed limits of...

I use FSD almost all the time, and think that with high likelihood it would have avoided the collision. If you were watching the visualization, you would notice that the blue "noodle" would project its path around the oncoming car. But I guess you wouldn't be looking at the screen.

In case a crime happens, and some witness says 'I saw a Tesla driving by', and police gets a subpoena on the Tesla's location data, an unlucky owner might get in trouble. Accidents or not, we don't want this to happen.

sunglasses?

I thought my explanation was pretty clear - in both cases I drive the car, let FSD do its thing, nudge the wheel when asked to, finally reach the destination and upon switching to "Park" see a dialog box mentioned above. FSD did not stop at any time, neither did I see a red steering wheel. When...

I always kept my hands on the wheel and got 2 phantom strikes over 3 months. At the end of the drive dialog pops up with something like 'Telemetry data showed inappropriate use...', without any warnings during the drive.

I don't use this setting, because from product perspective this is just a knob. Either positions would be considered 'normal'. Would Tesla engineers track how many people use it and when? Unlikely. On the other hand, disengagement count is not 'normal' (at least that's the way I feel). I...

I use it almost all the time, with a few exceptions. I'm on 11.4.4. One is heavy traffic. Despite 'Chill' setting, FSD always tries to change lanes to end up in the leftmost lane, which is suboptimal. Second is road work. FSD unsurprisingly ignores the red signs and tries to use blocked lane...

Just a bunch of cheap contactors would do the trick. Would have done…

Pedestrians are usually slow enough for the FSD to predict the safe slow-down, not even involving the AEB. Nevertheless, there are some exceptions where it can fail. I've had a person riding an electric unicycle jumping red light in front of me. FSD did not mark them as a threat (pedestrian...

Well, I'm kinda used to ADAS shenanigans, since I was using Comma Openpilot for several years. But this has been the hairiest situation so far. I had to quickly disengage and finish the turn, but, as this has caught me completely off guard, I couldn't send a disengagement report. Hope next time...

Here's the issue I observed 2 times over past 24 hours. Notably, this happened at the intersections that weren't previously seen as problematic. When making an unprotected left or right turn, FSD waits, commits to the turn and proceeds but then violently jerks steering wheel left/right and then...

In my 2 months of Autopilot experience, phantom braking is highly likely to occur when the camera detects traffic control devices that look like traffic lights. Two examples: 1. School zone warning lights that may or may not be blinking: Google Maps 2. Small yellow lights at the top of the...

increased following distance makes AV safer from rear-ending, but when it causes other cars to cut in and cars behind to overtake, increasing lateral crash risk for many others around, it turns it into zero-sum game at best

Looking at my leaking Samsung washing machine and fridge, I hope HW5 will be short lived

This is a good example of survivorship bias. Dolgov wouldn't share any clips where Waymo frankencar gets stuck in concrete. Tesla is exposed to ordinary people as testers who are happy to share their annoyance with FSD behavior.

The rules are, in turn, just generalizations of common sense, which can be represented as a tensor in a large enough neural network.

I have to push it all the time it's hesitating at intersections. Driving around with 11.4.4 on '23 MYLR HW4. Wheel oscillations got particularly wild today when the car saw a pedestrian at the corner talking on their phone and actively moving back and forth... apparently nascent AGI is...

Overconfidence is a well known phenomenon. If people were aware of how they actually look (i.e., as perceived by an average person) before dating, humanity would have gone extinct long ago. Iterative approach and conviction should solve FSD eventually.

Is this in relation to USS replacement when parking? Maybe there is some, but I don't trust these estimations because OCD wouldn't let me :)

AI allows all kinds of experiments that are not possible to learn while inside the car. For example, recording human drivers at busy intersections for multiple cycles and then flooding them with FSDs in simulations can help with planning.

Got 2023.26.11 on HW4 MY last night. No drastic differences/breakthroughs noticed compared to HW3. Same intersections present same challenges when making turns, unnecessary stop on blinking yellow, etc.

My few weeks old M3 has 2023.20.200, FSD 11.3.6

What would the driver do if the plan is wrong? Disengage early?

Driving is mostly muscle memory, no need to be Einstein

Guess where all the data and results are crystal clear? Mobileye powerpoints :) I briefly went through this paper: [2304.10410] Radar-Camera Fusion for Object Detection and Semantic Segmentation in Autonomous Driving: A Comprehensive Review Hope the quote below gives the taste how challenging...

As far as I know, doppler radar is not designed to detect stationary objects at all. Where do people get that confidence saying hi-res radar data is good? How about fusion being "well understood"? "Solved"? I tried googling for vision/radar fusion paper and the first one says radar...

4 weeks, took delivery late June

High-pitched sounds are usually produced by improperly shielded inverters/transformers. Plenty of youtube videos available with an explanation of the phenomena: https://www.youtube.com/results?search_query=inverter+high+pitched+nois

I don’t see a problem with vision-only L5 reducing speed to 5-10mph. Slowing down is less costly than radar/LiDAR.

I totally think some AV companies have reached higher levels of autonomy than Tesla, especially in good weather spots. But does anyone have an affordable solution? Should autonomous cars that cost twice as much as a good helicopter?

The radar backup seems to me to be the same trolley problem. A predefined confidence threshold, or a regression of some ambience factors, will be needed to confirm via radar. This may involve significant engineering effort. And additional hardware. Phantom braking seems to be an okay compromise...

I don't agree with the need for radar. While it may be useful for redundancy, but when all systems work, how can we safely combine the data? For example, a low-res radar w/Kalman filter can easily erase things like protruding metal bars from concrete paving. Is it safe to put equal weights...