Комментарии:
I don't want to say 'never' but I agree with you as the technology stands today.
ОтветитьSelf driving cars are safer and will make roads less congested only if 100% of traffic are robot drivers. Humans and robots sharing the road is the worst combination.
ОтветитьThe take I get from this is we're years away from robots going sentient and kill us all
ОтветитьI still think the whole approach is kinda stupid.
Currently everyone is trying to make only their car capable of self driving to become the most biggestest company. So every car needs to actually be able to identify and track everything going on around him.
To me, it would make far more sense to push for global standards and try to slowly prepare the infrastructure itself. So every new traffic signal, sign and car is required to have a small short-range radio that can broadcast some information about itself.
If you have that, there simply is no longer a need for a camera to figure out what a road sign says, the road sign can just tell the car what it says. Tracking other vehicles would likely still be done for safety but the computer could take the messages from the other car into consideration.
Obviously this approach isn‘t going to make you the richest tech company as it allows everyone to develop on these standards but it would probably give us better chances of actually working self driving in the next 30 years or so, when most things have implemented these new standards.
Waymo and cruise is tr@sh
ОтветитьThe only way self-driving would work is, if everything on the street would be controled by an AI. So, it will never work. There can't be anypedestrians, animals or anything else... so basicly a highway without human drivers. It was a intresting idea, but it failed.
ОтветитьYou cannot conclude that deaths per 100M miles are higher for Tesla by comparing 0.5 to 0.36 because 0.5 comes from only 15 events and has high sampling error. In Poisson approximation, we can estimate the error to 15 as sqrt(15)~4 which gives 4/3000=0.13 error for the rate, which is comparable to the difference between 0.5 and 0.36. Not to mention the error in the adjustment procedure that leads to 0.36
ОтветитьI don't think that our government should allow these companies to experiment on humans in himxn inhabited cities. If even one person is injured or killed by one of these vehicles, then they don't belong on our roads. I would never get in one.
Boeing has an auto-land system but there is nothing in front of it.
Pull these things off the roads.
Even if they get the tech hammered out, I will never ride in or use one of these self driving cars. I like the act of driving. Turning that over to a computer feels like losing an essential part of my experience as a human. You do you, but I'll be dead in the ground before a robot drives me anywhere.
ОтветитьThis video had been at the front of this channel for months, even though a lot of videos have made afterward. Initially I wonder why. However, after watching this video I understand the reason. This video is ultra important because it could save lives. And I'm sure it has saved several lives since being uploaded. Good job👏👏👏
ОтветитьIf our feral government wants autonomous cars badly enough, they will soon be all over the place, and the warnings in this video won't matter. I think we can expect many changes to roadways and diving regulations to accommodate the self-drivers. The harder the government pushes, the more costly in lives and taxes this will be.
Ответитьwaymo was started within Google
ОтветитьMy question is...
Why is this only affecting Tesla and Cruise and not waymo?
It wasnt me is the problem, YES IT WAS YOUR FAULT
ОтветитьAssume a child waiting at a crossing corner will dart out if they are not holding an adult's hand or stimy bouncing. Especially boys. Good luck teaching that to robocar 2000.
ОтветитьNever going to happen is just utter bs. I know a guy who did his phd in the field and he told me it’s actually done the only issue is the cost of lidar needs to fall rapidly and become standard and Volvo (geely) is exactly pushing that. Then the socs need to catchup to current tech which is 5 years of in highest end cars
ОтветитьThe problem is plain and simple - Computers can't 'think'; they just run programs.
'Thinking' isn't a calculation. It's a quality that's unique to biological brains.
AI can be fooled easily with just a fake traffic sign or a fake red traffic light. Someone can just hold up a painted stop sign or a painted red signal by a road & stop all self-driving cars there, and then may be rob passengers, since there is no driver & the self-driving car surrenders to a painted stop sign. AI is just a stupid thing, since it can't anticipate unknown problems, but humans can. The only problem AI can solve are the ones it has been programmed for. Even with its neural net, it can't understand something that it has never encountered before.
Weather forecasting - AI & super computers are good at, cuz weather patterns have been researched over many decades & programmed/modeled for already by humans to be fed into AI's neural net.
Complex data analysis - AI is great at, cuz AI can quickly look for patterns in a defined dataset.
Real world understanding - AI is bad at, cuz it can't model for infinite possible ways that a human can react with in a traffic situation, and truly speaking, humans have infinite ways to react to any situation, so AI can't prepare for everything. In fact, humans themselves fail when we meet other humans & can't predict everyone's behavior. So, it is stupid to expect AI to understand it all.
New Teslas do not have radar built in as stated in this video clip.
ОтветитьWho will be held responsible/liable when the AI technology fails and someone is hurt/killed?
ОтветитьI dunno if it'll never happen but i do know Telsa's pure camera based system isnt gonna make it.
ОтветитьWaymo works pretty well. Cruise doesn't. Unfair to blame the whole industry due to one company.
ОтветитьElectronics fail
ОтветитьSelf driving is for lazy brain dead people. Its like putting people in a prison cell for there own protection. I dont need self driving to be safe.
Also where is the fun of doing nothing in a powerful machine such as a car/ truck/ motorcycle? Driving is freedom, much like riding a horse but better. Driving is exciting and fun and good exercise for the brain.
I'll not trust an autonomous vehicle until it has something to lose - such as a family that it cares [sic] about or loved ones that it's anxious to get home to.
Saying "Humans make mistakes, so machines can be forgiven for making them too," is not good enough. If an autonomous vehicle isn't many magnitudes BETTER than a human then it's worse than useless. Worse, because its supposed abilities inspire false confidence.
I think about two million years of trial-and-error testing should see those algorithms fit for purpose. You know - like the anthropoid brains they're supposed to be replacing.
you need to put some tech under the road and sensors for self driving with safety!
ОтветитьOverall this video is decent but trying to criticize a self driving car for stopping when it doesn’t recognize a complex situation is unhinged. If anything human drivers should be doing this more. Wtf lol
ОтветитьAll this means is that the technology is not perfect, just like any technology, it will improve over time. You pick out specific incidents of where it failed, but those are the incidents that will be recorded and help the tech improve.
In general I agree of 10% of everything on your channnel. I subscribe to you mainly to hear the argument of the other side and see how people think over there lol
Self driving cars will work but only if every other car on the road is also self driving
ОтветитьThe conflicts with emergency vehicles is most definitely a data collection issue. They simply need to intentionally monitor emergency radio bands and send the autonomous cars (with an onboard human) to emergency scenes to collect 1000x the amount of data they currently have for training the autonomous cars how to respond to emergency vehicles. Problem solved.
ОтветитьThe next Fever brewing is Artificial General Intelligence. Like bruh, you can't make technology drive a car for you, you want it to think for you. Damn.
ОтветитьSelling a death trap for 200$/month. Dude, i want to to be Elon next time.
ОтветитьThere was a story in the media about Tesla Autopilot intentionally switching off if it detected an iminent collision This was to improve the crash statistics, as the vehicle would appear as human operated as opposed to autopilot operated at the exact moment of the crash and record the state of the vehicle as such. Can you please comment the veracity of this story?
ОтветитьIt's a lot to say self driving cars will never happen it's mindset like tht which would have us stuck in z stone age not a Tesla fan boy but self driving needs to happen thy probably need a different approach...
ОтветитьWho wants to sit in a car and watch it be drive. Id rather drive. And of course its more angerous, it introduces many new points of failure
ОтветитьIts still new tech . Obviously there will be bugs to sort out . Stop being so hyper critical
ОтветитьI don't think it's fair to put Tesla together with the others, as the others use lidar/sonar and thus are working on a very different problem on the technical side. Think like this, the tesla uses cameras and needs to calculate a distance to an object, it's only an estimate and it can be totally wrong… meanwhile the others just know the distance.
ОтветитьReal problem around self-driving is...AI is not really intelligent. Machine learning AI is actually a glorified statistics, so if there's no statistical relevance around current situation (roads, people, other cars, obstacles, etc.), self-driving go mad.
ОтветитьAre autonomous cars possible? Yes! However the cars would have to communicate with each other for that to work. The adaptation barier for that is obviously too high for it ever to occur.
ОтветитьI’m sure autopilot in planes had no issues when they were implemented. This video won’t age well with Tesla AI controlling their vehicles.😂
Ответитьself driving cars are not safe, period. There will never be a time when self driving cars are safe. It's literally the 21st century version of snake oil
ОтветитьSelf driving cars will be the last nail in the coffin of personal freedom.
ОтветитьAbsolute waste of money & utter scam/vapprware that's really there to WOW dummies with money (I mean ""investors"")
A REAL good use of that money can go towards actually developing better public transport. Reduce traffic on the road, reduce pollution, reduce roadway footprint, encourage people to interact with each other more, get a greater density of people to their destination sooner, since they're not driving, they can do tasks while in transport...
I mean, it's the obvious answer here, but no, everyone wants their own futuristic pod. People pay money to be lazier & lazier & more isolated from others.
you're an idiot if you think it's never going to be possible especially with all the growing technologies as evidence.
ОтветитьI've watched a few live streams of Waymo taxis and I was very surprised how advanced their system is. It's not perfect but it does look promising. Seems like they have the best software right now.
ОтветитьSelf-driving flying taxis are already in the air (carrying missiles). 🎉🎉🎉
Ответить"It will be like owning a horse in 3 years."
Sure it will, Elon.
I've long said that self-driving cars will need to be in a paradigm where they are not only driving the car and using GPS or visually seeing road signs and other vehicles, but are also in communication with one another, with infrastructure built into roadways, and even where needed a central traffic processing computer/A.I. In this situation there will be no intersection faux pas. In fact cars won't even stop at intersections and will seamlessly weave through with one another.
ОтветитьThis is not an impossible goal. Autonomous vehicles will become a reality and the norm in time. When the vast majority of cars on the road are autonomous (along with infrastructure designed for them), they will be FAR safer than human drivers.
Ответить