Thank you for the shout out! I appreciate it! You've inspired me to do an edge case video this weekend! I have a few reproducible issues that would be a good test for each version. The tunnel in my video is also a railroad bridge. There are two tunnels close together. I'm going to try going both directions through both bridges this weekend. I also use a 15% speed offset. I wish they would allow separate offsets per mode. I'd like to have Chill 0%, Standard 15%, and Hurry 40%. Then it would be easy to switch as needed based on situation. I have some areas that I really want to go exactly the speed limit.
The new system for the profiles on the highway seems to automatically go at, or above the speed limit depending on profile. Chill will stay at lower speeds, standard will be a little faster, and hurry goes much faster. When they come to city streets I imagine they will function similarly
Thank you! You should’ve seen me when I got the first email-I was completely stunned. I ran to tell my wife, and she just gave me that look, like, “Oh great, here we go, more Tesla talk.” A few hours later, the second email came in about another subscriber buying a car, and I was floored. I sat there, grinning like a kid on Christmas morning, thinking, “This is one of the nicest things anyone’s done to support the channel!” After posting over 100 videos, it’s the first time it really feels like this UA-cam dream might actually work out. Plus, it’s such a huge help with charging costs.
I've driven into NYC multiple times and Boston through that weird underground big dig using the FSD trial and it's 96% there. Yesterday morning I had it in hurry mode getting off the FDR at 61st in heavy traffic. The car started to scream because it couldn't handle the turn so I had to take over to set it in the correct turn lane, and when I put it back into FSD, it drove out of the correct lane and into the opposite side of the street haha, that was sick and I had to take over to rush to get me back onto the right side. Luckily there was no oncoming traffic. Right now it can't handle complex turns where four lanes of cars speeding off the FDR are trying to battle it out into two right hand turn lanes. I was also down in the West Village and the car RAN OVER a traffic cone in a pothole in the middle of Leroy St -- flattened it and hit the pothole so in these extra weird scenerios where it's not perfect. I didn't want to intervene because I needed to see what it would do and it didn't what I didn't want it to do. If you go into NYC, go down to 7th Ave and onto Leroy -- let us know if your car runs over the cone and pothole and whether they replaced to old cone haha. If you go into Manhattan, just let the car do it's thing. You already know it's somewhat weird traits but I can assure you it's pretty meticulous and safe with the occasional hiccup. I'm not going to lie, it's so much fun!!!
I hope us HW3 folk get the speed profiles before we get left in the dust (that would solve a lot of issues with using FSD in city streets; 2018 M3P with 123k miles here), but I think there's a misconception on the Auto Speed setting.. it's not necessarily meaning that the car will go, at all times, 15% or 40% above the speed limit it just means that FSD will still choose the appropriate speed based on surrounding traffic but if traffic is flowing and the offset allows it then it will go faster. And also depending on the Speed Profile (So Chill, with a 40%, might not ever get close to 40% over the limit.. but Hurry will be more inclined to get close to the max offset).
I’m a buy-and-hold kind of Investor, and that’s the only advice I’d give to anyone willing to listen. I don’t worry much about the share price-I focus on accumulating shares because, in the end, that’s what really matters. Honestly, I’m hoping to hit 1,000 subscribers soon so I can monetize my channel. My plan is to invest everything into TSLA and eventually use the gains to buy a Cybertruck when the stock price reaches a point where I can afford it. Essentially, thanks to you all, Tim Cook, and Elon, for helping me get my dream car. Right now, it’s just a dream, but we live in the greatest country on earth, so anything is possible! Recent events prove that.
@@FSDRoboRides Same here! Nobody is smart enough to time the market. But are you smart enough to know that you re not smart enough?! Personally i m hoping to be able to retire early (60 maybe?). I really don t like working xD I m a 31 yo italian, the economy sucks bad here, the government barely pays pensions as it is and I don t have a real profession. I m practically prostituting myself (not in the sex industry xD) to accumulate as much Tesla as i can. Tesla is my hope of having a decent leaving in the long term. Do your research boys!
I hope that's a compliment lol. I can tell you that in my opinion you are betting on the right horse. Im rooting for you. I feel confident Elon will do his best to help us achieve both of our goals.
You’re better than me, lol. I wish I could say the same! Back when I first started in 2019, I tried my hand at the price-guessing game, and let’s just say it cost me a few shares. After licking my wounds, I decided to stick to good old buy-and-hold, and honestly, it’s worked out great-especially these past few weeks. Now, my family thinks I’m a genius, but I know that won’t last long. They were begging me to sell at $108 because they couldn’t take the pain, but I held on, and I’m glad you did too. If we made it through that rollercoaster, we can definitely hang in there for a few more years to see what this abundant future Elon keeps promising will look like. P.S. I’m really hoping Tesla buys a stake in xAI-I think that company could be massive when it evolves into the everything app!
If FSD can truly 'crush' difficult scenarios, then FSD could drive in big cities during rush hour traffic, with zero human intervention required. FSD is nowhere even close to that level of competence.
Humans are not terrible drivers. Humans are actually excellent drivers! What you have to take into considerations is that there are A LOT of cars that are on the roads and they drive HUGE distances.
Thank you for sharing your opinion. Im just curious, what metric are you basing your claim on? Because 40k of death and 1 million injuries per year is not excellent in my opinion.
@@FSDRoboRides300.000.000 cars in USA. 3.200.000.000.000 miles driven per year, in USA. Drunk drivers included in this. And speeding cars. And unavoidable accidents (accidents that too will happen with a computer behind the wheel) Slippery, ice, snow, hail, heavy rain, dense fog,... Tesla FSD drives worse than a learner. FSD will drive off the road within minutes in winter conditions.
I get it, but it’s the only way to highlight that I’ve made over 100 videos, attracted 85k views, and still only had 500 subscribers until last week. After some self-promotion, I’m now at 775 subscribers. I promise to scale it back once I hit 1,000! So we can do some testing in NYC.
What do you mean? How else would the models be trained? Human input acts as a system of checks and balances. Right now, I believe Tesla runs FSD in shadow mode, compares human interventions to the system's suggestions, and uses that output data to train the models.
@@FSDRoboRidesAs I said. They use data from real life and simulations at then they adapt the model with that data and then create a new software version. So what you do in the car has no effect on that software version and underlying map version. It will never "learn" like a human learn each second you are driving. And most of the data Tesla can collect is never used in their training of the model.
I’m over here risking my life testing FSD, and only a handful of you found the subscribe button-come on, people! LOL.
Thank you for the shout out! I appreciate it!
You've inspired me to do an edge case video this weekend! I have a few reproducible issues that would be a good test for each version.
The tunnel in my video is also a railroad bridge. There are two tunnels close together. I'm going to try going both directions through both bridges this weekend.
I also use a 15% speed offset. I wish they would allow separate offsets per mode. I'd like to have Chill 0%, Standard 15%, and Hurry 40%. Then it would be easy to switch as needed based on situation. I have some areas that I really want to go exactly the speed limit.
My pleasure. Keep at it it only takes a few video to make it past that 1000 sub milestone.
The new system for the profiles on the highway seems to automatically go at, or above the speed limit depending on profile. Chill will stay at lower speeds, standard will be a little faster, and hurry goes much faster. When they come to city streets I imagine they will function similarly
I agree.. when there's one crash FSD has a "recall" (mandatory update), but there's never been a driver recall with thousands of deadly crashes
Great point. I wander what that Drive recall data would look like.
great video!
thank you so much. Hope to see you in the subscribers list.
Nice Job! Congratulations on the referral sales! Keep it up!
Thank you! You should’ve seen me when I got the first email-I was completely stunned. I ran to tell my wife, and she just gave me that look, like, “Oh great, here we go, more Tesla talk.” A few hours later, the second email came in about another subscriber buying a car, and I was floored. I sat there, grinning like a kid on Christmas morning, thinking, “This is one of the nicest things anyone’s done to support the channel!” After posting over 100 videos, it’s the first time it really feels like this UA-cam dream might actually work out. Plus, it’s such a huge help with charging costs.
Enjoy your free chargers Max! Thank you guys have a safe ride on your new Tesla 👏👏👏 @@FSDRoboRides
I've driven into NYC multiple times and Boston through that weird underground big dig using the FSD trial and it's 96% there. Yesterday morning I had it in hurry mode getting off the FDR at 61st in heavy traffic. The car started to scream because it couldn't handle the turn so I had to take over to set it in the correct turn lane, and when I put it back into FSD, it drove out of the correct lane and into the opposite side of the street haha, that was sick and I had to take over to rush to get me back onto the right side. Luckily there was no oncoming traffic. Right now it can't handle complex turns where four lanes of cars speeding off the FDR are trying to battle it out into two right hand turn lanes. I was also down in the West Village and the car RAN OVER a traffic cone in a pothole in the middle of Leroy St -- flattened it and hit the pothole so in these extra weird scenerios where it's not perfect. I didn't want to intervene because I needed to see what it would do and it didn't what I didn't want it to do. If you go into NYC, go down to 7th Ave and onto Leroy -- let us know if your car runs over the cone and pothole and whether they replaced to old cone haha. If you go into Manhattan, just let the car do it's thing. You already know it's somewhat weird traits but I can assure you it's pretty meticulous and safe with the occasional hiccup. I'm not going to lie, it's so much fun!!!
Thank you for sharing your experience. I look forward to testing this version in NYC. I appreciate your route suggestions.
96 percent is not good enough. Not even close.
A human driver is 99.99 percent competent.
Safe travels to you new Tesla owners...🎉🎉
Thank you so much. You are all the best.
I hope us HW3 folk get the speed profiles before we get left in the dust (that would solve a lot of issues with using FSD in city streets; 2018 M3P with 123k miles here), but I think there's a misconception on the Auto Speed setting.. it's not necessarily meaning that the car will go, at all times, 15% or 40% above the speed limit it just means that FSD will still choose the appropriate speed based on surrounding traffic but if traffic is flowing and the offset allows it then it will go faster. And also depending on the Speed Profile (So Chill, with a 40%, might not ever get close to 40% over the limit.. but Hurry will be more inclined to get close to the max offset).
I think your right on the money with these assumptions.
Great edge cases. Amazing drive choise! Do your best to report back to Tesla the mistakes and inaccuraces.
Do your research, invest in Tesla :)
I’m a buy-and-hold kind of Investor, and that’s the only advice I’d give to anyone willing to listen. I don’t worry much about the share price-I focus on accumulating shares because, in the end, that’s what really matters. Honestly, I’m hoping to hit 1,000 subscribers soon so I can monetize my channel. My plan is to invest everything into TSLA and eventually use the gains to buy a Cybertruck when the stock price reaches a point where I can afford it. Essentially, thanks to you all, Tim Cook, and Elon, for helping me get my dream car. Right now, it’s just a dream, but we live in the greatest country on earth, so anything is possible! Recent events prove that.
@@FSDRoboRides Same here! Nobody is smart enough to time the market. But are you smart enough to know that you re not smart enough?!
Personally i m hoping to be able to retire early (60 maybe?). I really don t like working xD I m a 31 yo italian, the economy sucks bad here, the government barely pays pensions as it is and I don t have a real profession. I m practically prostituting myself (not in the sex industry xD) to accumulate as much Tesla as i can. Tesla is my hope of having a decent leaving in the long term.
Do your research boys!
@@FSDRoboRides P.S. Bought about 150 times in the last 4 years. Never sold a share ;)
I hope that's a compliment lol. I can tell you that in my opinion you are betting on the right horse. Im rooting for you. I feel confident Elon will do his best to help us achieve both of our goals.
You’re better than me, lol. I wish I could say the same! Back when I first started in 2019, I tried my hand at the price-guessing game, and let’s just say it cost me a few shares. After licking my wounds, I decided to stick to good old buy-and-hold, and honestly, it’s worked out great-especially these past few weeks. Now, my family thinks I’m a genius, but I know that won’t last long. They were begging me to sell at $108 because they couldn’t take the pain, but I held on, and I’m glad you did too. If we made it through that rollercoaster, we can definitely hang in there for a few more years to see what this abundant future Elon keeps promising will look like.
P.S. I’m really hoping Tesla buys a stake in xAI-I think that company could be massive when it evolves into the everything app!
If FSD can truly 'crush' difficult scenarios, then FSD could drive in big cities during rush hour traffic, with zero human intervention required. FSD is nowhere even close to that level of competence.
Let’s get to a 1000 subscribers and find out.
Humans are not terrible drivers. Humans are actually excellent drivers! What you have to take into considerations is that there are A LOT of cars that are on the roads and they drive HUGE distances.
Thank you for sharing your opinion. Im just curious, what metric are you basing your claim on? Because 40k of death and 1 million injuries per year is not excellent in my opinion.
lol humans are awful drivers
@@FSDRoboRides300.000.000 cars in USA. 3.200.000.000.000 miles driven per year, in USA.
Drunk drivers included in this. And speeding cars. And unavoidable accidents (accidents that too will happen with a computer behind the wheel) Slippery, ice, snow, hail, heavy rain, dense fog,...
Tesla FSD drives worse than a learner.
FSD will drive off the road within minutes in winter conditions.
Dang. The self-promotion is a bit much.
I get it, but it’s the only way to highlight that I’ve made over 100 videos, attracted 85k views, and still only had 500 subscribers until last week. After some self-promotion, I’m now at 775 subscribers. I promise to scale it back once I hit 1,000! So we can do some testing in NYC.
FSD does NOT learn anything from human input. It all has to go into a new build, or into map data, or into both.
What do you mean? How else would the models be trained? Human input acts as a system of checks and balances. Right now, I believe Tesla runs FSD in shadow mode, compares human interventions to the system's suggestions, and uses that output data to train the models.
@@FSDRoboRidesAs I said. They use data from real life and simulations at then they adapt the model with that data and then create a new software version.
So what you do in the car has no effect on that software version and underlying map version.
It will never "learn" like a human learn each second you are driving.
And most of the data Tesla can collect is never used in their training of the model.
I can tell you’re in an inferior model y
It’s what I could afford. I hope you subscribed so I can upgrade to the inferior Cybertruck.