Guadaloopy

Never said it was unfair. I said it was engineered to produce a failure. The fact that so many complicating factors had to be introduced in order to generate the failure is an important piece of the story, IMO. My takeaway from the test is that Tesla has important work to do to address performance in these difficult edge cases. But the good news takeaway is that the system performs so well in normal situations that the critics ignore those scenarios.

100% agreed. It has to be monitored constantly, but even though you are monitoring it, you are still watching the car drive itself without your physical input. That’s a weird feeling at first. It used to feel really weird, because the maneuvers were mechanical and herky jerky - more digital if you will. As the system developed further and especially with the shift over to AI-trained V12, it is now much more analog and human-like with its maneuvers.

JV Day could provide the spark.

It’s a very uncomfortable feeling the first few times you use it. I don’t think that’s an indictment of the system, but a natural response that comes from trusting a machine to do something that you have had control of for your entire adult life.

That’s a three year old video, so it’s not an FSD Tesla in the video. It specifically says in the video that the initial speed is driver controlled and it is relying on the Pedestrian Automated Emergency Braking System to avoid the collision. It’s obvious that in this test, the non-production LIDAR assisted PAEB system outperformed the standard production PAEB system in older Teslas. Since the timeframe of the test in this video, Tesla has significantly improved its low-light image processing software. Couple that with the advances in FSD in the past three years, and I would be very interested to see a new comparison test between these technologies.

Not moving goalposts. Just replying with related experience. I do not have the specific experience of a child running in front of my vehicle while in FSD. I do have experience driving in my neighborhood and observing how the car sees and visualizes and behaves around the kids that play, skate, and ride bikes and scooters in the street.

I have driven around plenty of pedestrians and watched closely how the car reacts to them while in FSD. It’s almost maddeningly cautious. My experience has also been that the car sees and visualizes pedestrians on the display before I even notice them.

Pantsed? MrBig pointed out information that I had not noticed. I acknowledged it and I thanked him for it. It would be a better world if we could all admit that none of us always has the whole picture. I admit that I am biased based on what I experience consistently while using Tesla’s FSD system. I’m still a bit puzzled by that video because it doesn’t match my experience observing first-hand how this system operates. But I’m willing to admit that my experience isn’t definitive. Clearly, the majority of folks on this thread have decided that they have the whole picture regarding Tesla’s approach to FSD and have passed final judgment. They highlight information that supports that bias and dismiss any information that doesn’t. I’m willing to participate in this thread and be a bit of a punching bag because I think it’s important to share another perspective. But y’all tell me to fuck off and leave the thread, and I will.

Read the comment I was responding to. Pato was calling out a Level 2 system for behaving exactly like a Level 2 system is designed to behave.

You can’t have it both ways yet. If you want a level 2 system with human supervision, the human has to have the ultimate override. EAB systems in all auto manufacturers can be overridden by pressing the accelerator.

Except that isn’t at all what I said. I said that it was part of the test design intended to produce a specific result.

You are correct. I did not notice that before. Thanks for pointing it out.

I would fucking love to go back to 1994 for a visit.

Since the thread got necrobumped, I can go back and necro-rep a criminally underused movie quote.

Team is behind the pace of the famous "Tombstone Season". That team was 8-12 after 20 games.

It's honestly my only complaint about the car, and I had the exact same issue with a Mercedes GLC 300.

Pretty much already answered but it does the following: Verifies that all windows are fully raised Turns off automated windshield wipers Locks doors, trunk, and charge port Folds in side view mirrors Places HVAC system in recirc mode (see below)* Places car in neutral/free roll It's a handy feature to be able to hit one on screen button to verify all of these items. *Unfortunately, due a design flaw in the air intake location for the ventilation system in the Model 3, it's a very good idea if not an essential step that you put HVAC in recirc for a car wash or during the rain or you run the risk of ingesting water through the intake and soaking the filters and ending up with a mildewy funk in the HVAC system. Car Wash mode is one mitigation that Tesla developed to address this common complaint.

That's why I suspect that as in previous "tests" engineered by O'Dowd, the driver is actually applying pressure to the accelerator to override the automated actions of FSD. The behavior of the vehicle during this "experiment" is so far out of step with my actual experience with the system that I believe that the car is being manipulated by the driver.

Yes, FSD failed this particular test. It's actually a very difficult test. It doesn't take a very critical eye to watch this and see that the test was engineered to result in failure. - The dummy was dressed in neutral colors and earth tones to blend in with the background vehicle and background vegetation and wall. - The dummy remained stationary until very late in the test. - The dummy was accelerated to a near instantaneous running speed at the last possible moment. These conditions would be difficult for an unsuspecting human driver. O'Dowd is a disingenuous and untrustworthy Tesla FSD critic. He is a competitor, and he is not above manipulating and falsifying experiments to show the results that support his arguments. Previous videos from O'Dowd have showed a warning on the Tesla screen indicating that the accelerator is being pressed by the human operator to override FSD's automated actions. O'Dowd has passed these off as FSD failures when the failure is actually due to the overriding actions by the human operator. This particular video has a warning on the screen that appears to be unrelated to FSD, but if multiple warnings are active, you can cycle through them to see the one you want. There is something weird happening in this test. In the first three tests, you can see the vehicle start to decelerate as it approaches the crosswalk and sees something, but it doesn't stop. That leads me to believe that O'Dowd and his team are gaming the warning displays and actually slightly pressing the accelerator during these tests, effectively overriding FSD's automated actions. I would like to see this test completed on a vehicle with no visible warnings on the screen. I would also like to see videos from O'Dowd with the technology he supports passing this same test.

I may have been wrong.

Who didn't see that coming? FUCK!

Fuck this season.

Good job, guys. Solid swings.