Видео

I have model 3 highland. Same problem

HAhaha lucky you only for a day!! Yea I'm going to have to stuff something inside for my daily driver :)

If I don't drive much, should I charge to 60% and the next day maybe it goes down to 50%. Should I charge again to 60% or wait until it drops to 40% to charge again? I live in an apartment so can't be plugged in all the time.

I was the only one in a 25 mph with cars parked along side and the alarm went off. This is a total rip off for Tesla insurance. What can we do about this?

Is it on model Y or 3?

Turn off the music, get rid of the blurry box on the right. Your recording...why the music? Its louder than you talking.

👍

There is no value to using FSD for 5 minutes strolls through the neighborhood. It's slow to react and offers an overall janky, jilted driving experience. Since FSD is just a glorified LKAS/ACC system use it where it excels, long highway commutes on boring predictable routes.

After almost four years in beta, FSD cannot backup the car ( 34:56 ) . It also cannot understand common road signs: DO NOT ENTER, ROAD CLOSED, etc.

And I'm sure RUclips might have something to say about your music selection.

Peeps just don't seem to understand that you don't ALWAYS have to engage FSD when you are driving a Tesla. If you just feel like 'driving', go ahead and drive! FSD don't care. But even if you're driving without a fully engaged FSD, that smart ass Tesla is still providing some incredible safety features, whether it's the great 'blind-side' warnings or the proximity warnings that keep you from bumping into walls or other cars when parking, among others. And, IMHO, it ain't gonna be long before insurance companies will be rewarding drivers that use some kind of FSD...only makes sense. These peeps that are slow to seeing the wide ranging benefits of FSD are the same ones that thought smart phones were overkill when introduced.

I had something very similar happen. I was in 4-lane street split by double yellows in the very right. The car in front comes to a stop with the left lane empty, so FSD tries to go around the car. At the same time, the car in front tries to change lanes as well. Turns out there was another car street parking, which was blocking our original lane. Instead of stopping, FSD seems to try to jump the now moving car and double yellow while traffic is approaching. V12 is definitely more confident, but it can be a bit dangerous when doing things like this.

Great idea, but I think having one built by Tesla would be better. They could base it on the Performance for the superior 0-60 and handling instead of this Long Range model. They could also remove the pyro fuse that disables the car if the airbags are blown, so the car would keep going if it is damaged. It could be reinforced using outriggers from the giga castings to make it harder to disable. I don't think an aftermarket company would be able to do most of those things

It would be interesting to take the drive further and see what happens at the 10 Mile marker.

Nice edge case. It should definitely have started to break once it realized that the bus was moving.

Interesting case. Would have been interesting to see how FSD would handle it but it might just have been the awkward linger in the face of the opposing traffic. Or collision. Might have resolved it as the bus made way. It should have halted as the bus moved and let it pass and go behind it. On a side note, cost effective robotaxis could well be the death of buses so we get those awful boxes out of traffic. Trucking could also be smaller units with robotic drivers. What a breeze that would be.

Most states use green background with white Kettering for mile marker. May be your video can help counties to change the signs to green background.

Nice video about the auto park. When did you receive the 12.3.6 update? Tesla still has my 2020 at 11.1 with no updates. It is very buggy to say the least.

Should have been avle to do it in 1 go. Seems jist as bad as the last version of the feature

I think this is the first time I’ve seen Tesla’s reverse with full self driving. Very cool!

This is fascinating, but I really wish you people weren't beta-testing your crappy "FSD" cars on the same roads that the rest of us have to use.

Around me, it picks up route 55 as the speed limit in a 30 😅

It defnitly took its time to make sure...

its a tesla scam to increase premium

very frustrating

This happened to me yesterday and it was pretty amazing

Our stop signs are about 30 feet before the intersection. Causing FSD to stop twice, once for the stop sign and once for the intersection. Been needing to pedal tap to skip that first stop.

I’ve had the same problem on rte 20 and jct signs for rte 15.

pretty normal and correct response from the FSD, that car was going way too fast for that red light, and the FSD can see and process that the way we do, its a AI after all

My guess is it's not trained to recognize trash bags. If you look at the first time it goes around, it looks like a dog or deer appears on the screen for a split second. Then the second time it looks like a pedestrian. Presumably, it slows down to allow itself to process the image, and then if it can't recognize it as a reasonable probability of being some object it's trained to recognize, it doesn't regard it as an anomaly and keeps going. It would be an interesting software challenge - how would you program something relying on optical data alone to recognize an anomaly in the road? Maybe they could train a neural network based on roads with varied anomalies and roads without. I just don't know if that would be able to reach the required reliability.

The intersection and stop sign surely affected FSD’s performance. But the driver is always supposed to be in control so I’d say FSD handled it perfectly. No one would have been injured or damaged.

Excellent behavior, it stopped, determined the object was safe to run over and proceeded. Not sure why you wanted it to go around the bag. I can guarantee you Tesla FSD has seen more bags in the road than you and all of your ancestors have ever seen!

A very easy drive for a human, but FSD has problems: 1:00 Human must intervene. FSD made a big mistake 1:30 Human intervenes *again* . FSD can't make the Left turn 3:13 FSD drives over an unknown object in the road. Twice. Not good

Death. Trap.

Pretty sure there are no speeds markers that end in 1,2,3,4 or 6,7,8,9. So the system problem ignores everything but 0 and 5.

Thanks for doing these tests. Looks like FSD is learning.

Find a nice place without stop signs and things like that, and put different objects on the road. Daytime, night time, rain, sun, shadows and more, and test different speeds.

Teslas must have been trained by now on tens of thousands of obstructions in the road. This includes light empty bags which generally move with the wind and can be driven through without incident, whereas something static usually means it has contents which could be rocks, concrete etc and should be avoided, depending on size. I think, like other people here, that it was the combination of obstruction and being near to the stop sign that changed the behaviour. Ego proved it could avoid the bag, but again when creeping up on it (and obstruction not moving) FSD might have weighed probabilities and concluded it was small enough (and a plastic bag) that it could go over. Maybe I'm reaching, but its a possible scenario.

why is there mile markers every mile? was someone at the sign factory bored? where do you see 5mph speed signs? lowest i seen was 15mph in residential towns near children playing signs. setting speed based on signs isn't accurate. some areas dont have reduced speed zone end signs. so i occasionally see someone going town speed miles past the opposing lanes 35mph sign. so i just pass them. some school zones have electronic LED signs that change speed depending on time of day. can it read those even though its sparse led dots? likely not. who ever decided to code signs speed wasn't thinking hard enough.

what if there is a big rock in the bag 😊

wouldn't surprise me if the AI is keyed to seeing a '5' as an indicator for speed, considering almost all speed markers in the US have a 5 at the end.

why did you torture the poor AI?

It's not cheating, this is how you train the AI models.

So if the bag was a small animal? SPLAT.

테슬라 차주들이 다양한 데이터를 공급...어떤 회사도 이런 충성심 없다.