Why self-driving cars have stalled | It's Complicated
HTML-код
- Опубликовано: 7 сен 2022
- Fully fledged self-driving technology appears to be perpetually just around the corner. It is a promise that the Tesla chief executive, Elon Musk, has made almost every year since 2013.
Subscribe to The Guardian on RUclips ► bit.ly/subscribegdn
But in the real world, it is still an open question whether level five self-driving automation is actually possible. Josh Toussaint-Strauss finds out how close industry frontrunners have come to full autopilot mode and assess the scale of the problems standing in their way.
Driven to distraction: how close are we to watching films in self-driving cars? ► www.theguardian.com/technolog...
How self-driving cars got stuck in the slow lane ► www.theguardian.com/technolog...
The Guardian publishes independent journalism, made possible by supporters. Contribute to The Guardian today ► bit.ly/3biVfwh
Sign up to the Guardian's free new daily newsletter, First Edition ► theguardian.com/first-edition
Website ► www.theguardian.com
Facebook ► / theguardian
Twitter ► / guardian
Instagram ► / guardian
The Guardian on RUclips:
Guardian News ► bit.ly/guardiannewssubs
Guardian Australia ► bit.ly/guardianaussubs
Guardian Football ► bit.ly/gdnfootballsubs
Guardian Sport ► bit.ly/gdnsportsubs
Guardian Live ► bit.ly/guardianlivesubs
#selfdriving #cars #tesla #elonmusk #waymo 
Наука
Driven to distraction: how close are we to watching films in self-driving cars? ► www.theguardian.com/technology/2022/apr/20/driven-to-distraction-how-close-are-we-to-watching-films-in-self-driving-cars
Please explain your use of YEARS old video to claim that TODAY'S FSD is "NO CLOSER"
True self driving tech can not depend on cars alone. The tech also needs to be implemented into the Infrastructure like, markers or sensors on the road and street lights that can talk to cars.
Google/waymo already stated there finished with the logistics of there system the complications is the mass adoption of the standers they feel acceptable and waymo is in 30 states and 6 countries gave waymo full adoption support basically so to say iscomplicated is just dumb
Waymo self driving cars are here and real
@@roxaskinghearts except Elons fsd is better. Teslas dont all get lost down some dead rd all at once.
@@GWAForUTBE no waymo is better then even that and people are starting to understand the power of computer scientists hitting a project that are kept up to date compared to git hub quality vs quantity can benefit times more when the quality has tools like what these guys have they basically from the ground up made everything in waymo so its just learning how to better adapt this into a black box that is easy to install which i mean they could hire a car company which they have and then its just finalizing a car
waymo went software car which is so google
(Original video on my channel)
So that's actually my dash camera video from 0:54 to 0:55 and my Tesla was NOT in any autopilot at all, it had performed some emergency braking while I was driving manually. The traffic lights were out at the intersection and the other car just blew on through.
So the tesla actually saved you? Interesting
You upload everything but not this dash video?
Can you upload the original video on your channel?
@@logitech4873 yeah, sure why not I'll get that up there
@@logitech4873 it's uploaded to my channel now
Well, there's always another next year, Elon 🌝.
Lol🤣🤣🤣
And the year after that.
Elon love harassing his employees he ain't got time
can somebody explain what AI is??
@DylanNicolasSG What does this even mean? What "impossible" thing did he achieve? You know Lexus had adaptive cruise control 30 years ago right?
The greatest power human beings have is 'unpredictability'
Or predicting unpredictability.
Driving a modern Volvo and having it slam the brakes on me passing between two parked cars was all I needed to know about how "advanced" it is.
Logical conclusions are not your strongest skill I see
Volvo sucks. It's a Chinese car now.
One thing is to make autonomous cars driving around in sunny California. Another thing is to do the same in parts of the world with heavy snow, poor or missing road markings, and in a climate where you are nearer to the elements.
Tesla autopilot works even in heavy rain at night, roads with poor markings, and during snowfall.
Greetings from someone who lives north of the arctic circle.
@@logitech4873 except it doesn't work and needs assistance.
This is actually so true I live in LA but born in Cleveland I wouldn’t trust my self drive there
@@logitech4873 and cost like a zillion dollars. Who, besides the very wealthy, can afford it.
@@davidberchtold3859 They're not that expensive. I could afford a model 3.
the problem is if it even makes 1 mistake its extremely scary to the point where ppl wont try it again for a very very very very long time
I rather be full control then car controlling it self. I don’t even trust cruise control
Some solutions has to be from local governments... like tag to the machine see more details about the road... some streets has to have some cameras to help see strange situations.
Then you have at least level 3 autonomous with real safety.
Never tried it, heard scary stories.
Will *NEVER* activate it!!
I have a self driving car: The rail guided car (AKA the train)
You own a train?
@@RandomGuy-nm6bm No, I have access to one. Thought that was clear.
I own a self driving horse
@@KCJbomberFTW Electric or gas powered?
Is automated rail troubled by any problems with ai errors?
Public transportation like the metro and tram lines are kind of autonomous and electric!
this i'll never forgive elon for his BS hyperloop proposal which just ended up stalling CA from spending money on high speed rail
@Matthew Morycinski One would think that but. NOPE and NOPE
@@Xx_Eric_was_Here_xX Citation badly needed.
Because the actual history is that NIMBYs sued and red-county contractors scammed and that's where all the money went.
No California state money went towards a tunneled or lowered air pressure tube system.
elon is scared of poor people thus why he didn't like it
@@CrissaKentavr no money went to it because it was never even properly proposed. just a complete pipe-dream idea. if California is losing their money to their red districts that's their own problem for having garbage legislators. it is not missed on me how incompetent and bureaucratic the state of california is. but elon is also a 17 year old.
5:17 no matter how many "safety features" you put in your car, prevention is always better because the safety systems can fail just like what you see here.
like reducing the amount of car overall by making neigbourhouds more easy and safe for walkers and cycles
and improving the public transport network
I see self driving cars as a anti human solution since it keep humans seperate (in groups of 5 in a car) rather than making commuting more social like cycles can
@@comrademartinofrappuccino as an introvert, I hate "social circles" but I do see your point on the rest.
My only issue on commuting is that it's difficult to commute in my country, especially if you're carrying groceries. You can't buy as much as if you had a car as you have to consider hand carrying it on a train/bus or whatever unique point to point transportation you have in your country.
That isn’t a valid argument in support of cars though. You simply have issues with any other method of transport because of poor city design and infrastructure. I’d recommend checking out the channel ‘Not just bikes’ or just briefly looking into city design in the Netherlands if you’re interested.
@@triadwarfare I see self-driving cars as trying to optimise an inefficient problem. The most optimal form of transportation is actually bikes. They are far safer and allows urban planners to more efficiently put everything closer such that any place can be reached with either bikes or public transportation.
Not Just Bikes video, "Why Grocery Shopping is better in Amsterdam" descibes how grocery shopping in a bike-centric environment is not that big of a hassle compared to shopping the US.
@@swankykoala4836 Nope. Bikes are terrible. You get wet and cold if the weather is bad and you get skin damage from the sun. You also get lots of accidents from falling over. And the bikes get stolen all the time.
5:03 "This is what can happen under level 2 automation"
This can happen under level 3 automation as well. At level 1, 2 and 3 there needs to be driver attention, something that was missing here.
i think level 3 guarantees safety and road rules compliance, but requires driver to be ready to take control in reasonable time for him to get situation awareness (5 sec?) in case car got itself into situation where it can't do it job (i.e. road construction, weaver change, difficult road situation (in this case car slowdowns or stops completely until driver ready to take control, obviously to not create traffic jams driver needs to be ready to take control in reasonable time) , in level 2 driver is responsible for safety and he must maintain situation awareness all the time(no sms/book reading, or tv watching), level 4 means car is ready to be on the streets without human driver at all. (but human driver can be remote, but if there is no connection and road situation is too difficult car should be able to park itself without breaking road rules).
With level 2 no car company is liable. With level 3, e.g. Mercedes is liable for the actions taken by the system. You as a driver only have to be able to take over in a certain timespan 5 to 10 sec I think. With level 2 you as a driver always have to supervise the car's surroundings and the system's actions at all time.
And it can happen at level 0 as well. You let 1000 human driven cars and 1000 cars at level 3 do this test, I guarantee humans lose (given we have clinical similar circumstances).
I’m an engineer in one of those companies. I can tell you this their designs are flawed.
They just patch it up hopefully it may magically work out.
It's just impossible in India.
Even humans cannot predict what will happen next on our roads😂
I’ve been saying that to my Tesla fanboy friends for years. „Self driving in Nevada“ means nothing when we are talking about rain, snow, traffic in Paris or Rome. I guess in India the car’s brains would just 🤯
Are you really comparing your speed to a computer?? Fsd will save lives the world all over. Allow the elderly, the disabled and even the drunks drive wherever they need to go.
FSD is going to happen, im betting on Elon.
Long live the King of Tech
Elon Musk is a gombeen man
@@oscarosullivan4513 Elon is a brilliant engineer.. but what he does what so many can not is be such an amazing businessman. Yet he lives in a 50k boxable. What is your real gripe. Afraid you will miss the vroom vroom of your iceage car
@@GWAForUTBE Because he doesn’t understand effective transport solutions. Public transport, proper cycling and walking infrastructure is the best way to go.
This is waaaaay into the future. Automate the trains which are on rails then think of automating cars.
Lol... until the billionaire and wannabe billionaire con artists figure out how they can get wealthy/ier off trains you can just rule that out, sadly.
Malaysia's had that for quite a while
Alot of Countries outside the US have automated trains 🤓
Yeah... It's a great idea. I wonder why the self appointed genius billionaire Elon Musk never think about it
@@kaptenhiu5623 simple. More cars on the road than trains on rails meaning more customers. If fixing poverty was lucrative than he would have fixed it
this why we should know how our brain is really a great blessing
Keep your hands on the wheel no matter until it’s actually proven that the world has reached that capability. I do it with my Honda but still have my eyes on the road and on the wheel. Don’t be cocky or else you’ll end up being one of those fatal.
Which on giving you a better self driving experience the honda or tesla ??
So what's the point in self driving vehicles???
@@yoobrynner9492 There aren't any. Hence keep your hands on the wheel.
Won't leaving your hands on the wheel deactivate the auto pilot though ?
@@yoobrynner9492 What's the point in cruise control? It's convenience. It's more relaxing if you just have to concentrate on the traffic and potential dangerous situations and don't have to think about acceleration, about where to turn, etc. Also you have improved safety, as you have the full safety capability of a human plus the capability of the system, which can look into all directions at the same time, which doesn't get tired nor distracted.
It may be able to work in a place like Phoenix Arizona, but I can't see it happening ever in places like Europe or Asia where the streets are so much more varied, busy and narrow.
Yes, too many variables. Maybe on some closed-loop deterministic route for delivery vehicles or public transport or trains. But not mass, open-world driving. They cannot program the whole world and understanding of it into these computers. True self driving is probably indistinguishable from AGI.
You don't get it. That's in elon-time, which ticks about seven times slower.
It's like how fusion power is always 20 years away, except it's always next year.
The question seems to depend mostly on the threshold we collectively set for this technology to be considered safe. Is it "better than humans" or "virtually infallible" ? The first one might be achieved, not the second one
Both can't be achieved
If "AI driver" is not safer than human driver, then what's the point of an AI DRIVER then? Then there's no advantage of AI here. In economic terms, It's much cheaper to hire someone to drive for you in a regular car
@@kaptenhiu5623 I dont understand your comment, are you adding to to comment your are responding to?
The whole point of the film is that automation is not nearly as capable as a human driver. So how do you propose that a system better than a human could be achieved?
@@Imp5011 I was answering a specific point of the video that showcases some fatal accidents involving Tesla's "autopilot", and from there concludes that AI driving "already failed". When actually, for a very specific task like driving on well-maintained highway for several hours, AI driving is definitely safer than human driving. Cherry-picked examples of accidents occuring on a highway do not disprove that.
Is a versatile AI driver, capable of being safer than a human in city centres, country roads, roadwork zones, etc. possible? Maybe, maybe not. Tesla's approach seems to have reached its limits, but the lidar-based approach is still showing progress
A few weeks ago I was in a two lane and came up on a dodge pickup in the leftmost lane following another small car too closely. I was in the right lane. I could see that the small car’s driver seemed distracted looking for where to turn. I slowed down because I figured the pickup might swerve suddenly into my lane if the car stopped quickly when it found the correct street to turn on. This is exactly what happened and I would have been in a collision had I not recognized the situation. I think this sort of thing happens much more than carmakers realize.
An autonomous car would just keep the distance. So it would also have slowed down, because it was too fast and too close. So when the other car breaks and turns, the autonomous car just also breaks and then accelerates again. Doesn't seem to be a problematic situation.
@@Duconi I didn’t clarify… I was in the right lane, they were in the left. That’s why I was able to see the person in the car. So I wasn’t in any danger technically (there were no cars ahead of me in my lane). However I knew what would happen if the small car were to stop suddenly. I slowed to distance myself from the situation. What I’m saying is I think it’s difficult to imagine an autonomous system recognizing subtleties like this. I always try to take a quick glance at the drivers around me to see if they might be looking at a phone, or distracted in any number of ways.
@@Mothball_man Well, if they would use autonomous cars, the car is not distracted by a phone. It can see all around all the time and doesn't get tired. So autonomous cars improve such situations.
Human reaction is sometimes unpredictable, true. But they implement the systems to drive in a very safe way. And if that driver is on another lane, it would be their fault driving into you anyway. They also don't want that and pay attention. You can not always predict that. So even with the best psychological skills, cars can crash into you. The advantage of autonomous vehicles is, that they can see all around the car and calculate the trajectory of other cars. So they react before you even know what they react to.
But why not combining both? There are some assistant systems, where you can intervene and have to pay attention. So you can get the best of both worlds.
By the way, the best way to reduce car accidents is to build cities in a way, that you don't need cars. With just a fraction of the cars on the road and safe ways away from cars for people, there are much less accidents. We have the knowledge for that for decades. I can recommend the YT channel Not just bikes for that.
@@DuconiI don’t think we are quite on the same page here. Of course it would be safer if that distracted driver and the pickup following too closely were both using autonomous vehicles. Yes it would be his/her fault. Yes it would be better if there were no drivers at all and there was a better system of mass transport. Yes.
But I am speaking of the here and now… as things stand. As it is now, autonomous vehicles cannot identify subtle indications of driver distraction. They are not capable of imagining possible scenarios, vectoring those possibilities, and creating a safer margin based on something that might happen based on said subtleties. As Elon said in an interview recently… “turns out making cars drive themselves is REALLY hard.”
It’s hard because the human factor, both in sensing possible mishap and causing mishap, is infinitely difficult to grasp without using the miracle of human gray matter.
@@Mothball_man the autonomous car can still react faster than you. I agree that it can not predict distraction of drivers. But what makes it really hard is not the safety of that situations but having difficult situations like construction zones, cars parking in the second row, trash on the street, handling emergency vehicles, etc.
The problem is not just seeing what's in front of you...but predicting what will happen in the next few seconds. As much psychological as computational.
I don't see a problem with that. The autonomous car can react much faster and can calculate distances much more precisely. In the end cars are just objects moving with a specific speed. There are physical laws on what could happen. The autonomous car can just drive in a way that under no circumstances it crashes into something (except another car crashes into the autonomous car).
Optimizing traffic and car population levels along with mass investment in public transport > making rich people/corporations even richer.
self-driving cars are the wrong solution. What we need is to re-design cities to be walkable and fully served by public transport. Self-driving is a bad solution provided by corporations to make themselves richer at our expense and safety.
No matter how much technology you pile onto cars, nothing is greener than being able to use your own two feet to walk to the store and get your groceries
Unless you physically move millions of homes, buildings, companies, stores, etc most of us can't and never will use public transportation. That's a privilege for rich folks in cities. The denser the housing the more expensive it is. I like the idea of public transportation buts it's a pipe dream for millions of us who can't afford to pickup our entire lives to move to a dense concrete jungle.
@@baronvonjo1929 absolutely, cars will remain the best mode of transportation in less populated areas. I was referring to transport within cities while attacking the “solutions” given by people like Elon Musk and auto-makers
@@baronvonjo1929 We are specifically talking about within metro areas. You can also run buses (with longer headways) to the centres of rural towns, since rural does not mean that a community cannot be walkable. The people who still prefer single-family homes can drive, and people in the middle of nowhere will still drive.
Why they removed the dislike button.
"We are no closer to autonomous driving," that's not true.
Some of what they showed as ai failing was manual driving
They said everyone paid $12k for autopilot
That's not true.
to solve the issue of things the car might not be able to see they need to use special cameras every few meters that would be placed on the roads so then the car would be able to get info of unseen objects by proximity to these cameras, so then like 2-3 cameras that are close to the car would be sending info needed for the car to avoid crash with unseen objects that are obscured by another object (could be even another car)
Waymo is at level 4 but Guardian conveniently left that out. Seems like a bit of a skewed report.
I've been selling cars for 20 years, I can tell you something. If you are dealing with self-driving technology, it's been out for a few years, it's definitely glitchy because it's basically cameras that are reading lines on the road. There are so many variables. But up here in canada, when it's minus 40, strange things happen to sensors, also strange things happened in a blinding white out blizzard!
That's not self driving. That's lane assist. A level 2 feature.
@@HenriZwols yes that's exactly what I'm saying if you can't even get that to work then what
@@HenriZwols Tesla said admitted their cars are level 2 and with the current hardware will never reach level 2. i personally would love to use an autonomous car, but a hood level 2 car and a driver who knows and respects the limits of the system make a save and pleasant driving experience for everyone involved. Even if tesla wouldn’t (and wasn’t, i went with Mercedes and VW) be my choice of car.. 😁😉
Self drive car, self pilot aircrafts....we will get there. Science always evolves. 500 years ago who would have thought there would be smoke alarm😂
Airplanes fly on autopilot for like 90% of the time. Basically only the taking off and landing is done by the pilots. And even then landing is helped with a lot of computing power
@@MrJimheeren yes, and I'm guessing the technology to autopilot a plane is a lot simpler than that needed for a car. There would be simply less objects to identify and far less to hit.
@@MrJimheeren What you don’t see is Autopilot for aircrafts is treated as a huge driving aid. Pilot’s have heavy workload and we have to multitask, keep in constant touch with ATC, and require to adjust through every info given such as change in weather, altitude, air traffic, etc. Those constant changes are then applied to the AP system and if the situation requires us to take full control, we take full control. Applying that to a grounded vehicle is not the same as supposed to flying. Your room is limited and you have no idea what is going through the minds of other drivers. The closest I’ll probably get to self driving is staying on cruise control on the freeway.
Who would’ve thought there would be smart phones?
Every single automaker and autonomous company has misjudged how long it would take to have autonomous cars. Projections change every year, just like Tesla. However, these misprojections are not newsworthy unless they are Tesla. Machine learning progress is hard to predict. Look at protein folding, Go, and other projects. People either overestimated or underestimated, and then suddenly they were here.
Elon does need to shut up though
Aren't self driving cars already on the road from last year, or even the year before?
Yes. There have been self-driving cars driving for for-profit companies in closed betas for about four years. In 2020 they were allowed to be open to the public.
It's limited areas, and limited services, but they do exist, and are constantly expanding.
Its all about saving people from being hit by cars or those who die in a crash. Already a Tesla on Autopilot is 10x less likely to get in a crash then the average human here in the USA. That the actual statistics from over a billion miles of real world driving with autopilot over the past 8 years.
i don't know how tesla came up with the values but i can say it has to be a false statement. i couldn't find any numbers how many times the driver had to take over the control when the autopilot made a mistake or disabled itself, when i see the FSD videos the value can't be zero. did it work at fog, rain, snow, night? does it work on every street? when the system works only on a autobahn then it needs to be compaired with the safty statistic for autobahn and not with a median of everything.
I guess it can never crash if it always disengages right before.
We could easily solve traffic and reduce car accidents by limiting cars in the city and improving public communication but it seems like we prefer a more expensive, dangerous and innefecctive solution instead that's just making things worse and grifters richer
What does that even mean? Who decides who gets to drive and who must be relegated to 'public communication'?
Translation...I don't drive into the city, so I recommend banning it for others.
@@candorsspot2775 at the moment... individuals decide based on traffic and parking. I guess democracy (with all its coercive power) decides to increase throughput through allocate some road space to buses (though this may decrease traffic)
@@Robert-cu9bm Learn how to read then
Self-driving vehicles could actually make public transportation more attractive but anti-cars/« urban planners » are too narrow-minded
We’re no closer to self driving cause we have to wait next year
Major errors in this video. The accidents with Teslas, equipped with so called "autonomous driving systems", had nothing to do with FSD, which is what the video is all about. The accidents was also not necessarily caused by the AP. Statistics shows that their software reduces the number of accidents significantly. Also, Teslas autonomous capability is way beyond competitors technology-wise, but according to laws and regulation it is to be treated as level 2.
1. so, the first 30 sec. you are saying he is just BSing and stall the conversation. 2. what is the conclusion? there is no closing remark on this video. just saying most accidents are fatal. did anyone review the script before airing?
My problem with this sort of analysis is that it never brings up the dangers of human drivers. Does the technology need to be perfect, or just better than human drivers?
Well right now it's making human drivers worse because the driver will only drive when something goes wrong.
What makes me pause is the surprises that require a split second reaction like a having a horse or deer in the middle of the road instant reaction maybe required. Even the best computers and smart devices are susceptible to spyware viruses and male ware those issues noticeably slow down the processing speed when executing tasks. An extra half second when it comes to loading a website or opening an app is of no consequence however the same half second latency when your going 70 miles an hour in a congested or hazardous environment could mean the differences between colliding into an overturned 18 wheeler and just barely missing it. Would be nice to jump into autonomous car and take a nap, watch tv or spend quality time with your family while your being driven to the beach on a family vacation. That would be an awesome achievement however at least for me it would come with some serious anxiety that would make it impossible to completely be at ease or able to relax.
You guys failed to do some research. @ 4:02 you start talking about the cow thing. Teslas new updtate include something called occupancy networks and occupancy maps, which detects every voxel of space and it’s velocity, to create a more redundant system and not have to rely on classification.
Just goes to show how powerful people's on board computer (mind) is
Not powerful, we just have years and years of real world data to identify possible obstructions, while computers have only had a few years. For example we know not to hit a cow or a sheep. A computer still has to learn that.
Or how well we use our senses
You have to think about every single cities type of street signs, pedestrian crossings, construction zone props, just every single scenario and road structure that’s different city to city and state to state.
I don’t see there being fully autonomous cars unless there are more standard and universal road signs (that are kept up and not overgrown with vines or knocked down), universal road lines, pedestrian and bike lanes, parking, and even installing sensors in these objects for the autonomous cars to communicate with or something like for special situations for unique parking lots or very busy cities. It’s all just too varied and too much to program and code.
The issue is we are trying to replace drivers with computers instead of attempting to enhance safety and awareness with automation. When you advertise self driving to the general public, without the requisite training or information, you are advertising the same general population who for the most part do not even bother to read their owner's manual of their vehicle, let alone even bother to be safe on the road. We need to focus on improving safety culture and education in driving then supplement it with automation.
With legal issues and trust issues this will never happen
If they were all tanks, self driving cars would have more flexibility. When they made mistakes they could just smash through everything else. Success!
“A group of robo-taxis malfunctioned and blocked an intersection-“
Yeaa that’s not suss at all
These were obvious problems right from the start
when I was a kid I thought that cancer would be curable by now. While advances are many and fast some thing are so difficult that seems to stall at a point
Less complex with Ai coming. Could be easier if they just installed installed sensors in the road aswell
I can see it causing more issues for companies with loss of life particularly in public being too burdensome.
Once this is tied to the social score could the cars be hacked to recognize and run over a transgressor?
The over simplification in this video is seriously jaw dropping. While many of the problems identified are genuine, the conclusions are so over simplified that this video is almost disinformation.
how can big name newspapers can be so wrong ?!
What if your driverless car hits and kills someone? Are you criminally responsible? Or does the car company get hit with a lawsuit? Seems like a lot of liability issues
This is completely out of date and absolutely loaded with misleading and distorted information.
@@darksam1212 Strawman argument. He didn't say this.
@@darksam1212 "I bet you think musk is a founder of Tesla,"
This is a very presumptuous guess, serving as the basis of the argument that comes next:
"so you are one to speak about distorted information."
This is a classic strawman argument attempting to convey that Ken is acting hypocritical, even though he never did say anything about Elon's status as supposed Tesla founder.
Not enough credit is given to Waymo in this video. I believe they are ahead of everyone else.
They really aren't. It's a geofenced system.
@@logitech4873 How does geofencing limit any of the everyday bustle or complexity in a given area? You clearly don't know anything
@@Telencephelon It merges pre-existing data from the geofenced area with the real time information to give it an advantage. Outside that area the system won't work. That's not necessarily a bad thing, but most people will expect anything described as self driving or fully autonomous as being fully functional anywhere anytime as well as a human.
@@msmith3395 After having seen Teslas AI Day 2 I doubt anyone is further. Most other companies are tackling geofenced AI with very high quality data and tests.
So are right. Tesla will stand out. The question is which will be better in the end - since most people really stay "fenced
as well and need FSD for commutes
@@logitech4873 And that is exactly why they are ahead of others.
This is a great video. I´m a journalist from Denmark, making a documentary on a.i., where im also explaining the use of a.i-tech in self driving cars. Would it be ok if used a small part of this video in my explainer? Specifically 0:51 - 1:04.
Kind regards!
Self driving car sounds cool. But reality very scary. Cool concept. Scary idea
In regard to traffic lights, they need to not rely on visual information that the AI might get misleaded by so instead of visual input of the traffic light it would get a fully accurate and reliable simple info data (sent to the car) of true/false of go/stop from the traffic light...that way such issue of car driving at red light would cease to exist, also to solve the issue of things the car might not be able to see they need to use special cameras every few meters that would be placed on the roads so then the car would be able to get info of unseen objects by proximity to these cameras, so then like 2-3 cameras that are close to the car would be sending info needed for the car to avoid crash with unseen objects that are obscured by another object (could be even another car).
They should invest in automated flying cars for personnel use. You don't have to worry so much about hitting things on the road.
YeahI want a hover car from Fifth Element have a orange head fall into my car lol but seriously agreed.
This video is so Tesla centric. There is barely any information on where Waymo is at, maybe because there have not been any huge failures all over the news. The information about Cruise is centered on their latest failure. No real investigation on "how close industry frontrunners have come to full automation". Anyone could have made the same video by just scouring the Internet, I expected better from the Guardian.
There's a simple reason WHY....
Humans are unpredictable.
A machine may take the perfect decision based on observed behavior. But it may still fail because humans could take an irrational decision.
No machine can predict illogical, irrational, and many a time, intentionally detrimental decisions.
Unless EVERY vehicle is autonomous, and living beings outside these vehicles are banned from stepping on the roads, Fully Self Driving will be next to impossible.
Agree completely. There's a lot more to driving than keeping to the road. You may need to differentiate between elderly and young people which in itself isn't easy for computers. An electric car will need to power not only the motor, aircon, lights and wipers but power at least two or preferably three powerful computers and relays motors etc for other essentials.
They don't have to be perfect. They just have to be better than us.
Yes, they did say it; it's called real AI - it's the same reason that actual AI is a laughable idea: many people mistakenly believe that is "intelligence" equals "logical reasoning", it does not, making calculations quickly does not make you intelligent
E.M. is a supervillain in disguise 😂
I find this re-assuring.
I think the main reason why they stopped developing a fully automated vehicle is the immense scale of potential job losses and the resulting disruption of social order. What would happen to all the millions of unemployed drivers? They need to be occupied/controlled to have order in society, and not many of those people can be re-integrated to the workforce. So, in order not to disrupt the fabric of society, the Big Tech has abandoned the project for now.
Like Fusion, complex engineering challenges don’t reveal true depths of complexity’s until significant capital is committed and countless failures are overcome. Perhaps the question to ask is why Tesla is allowed to profit selling Hope of successful software and why the Beta lab they are permitted to operate within (Real roads with unwilling / unknowing participants) is an acceptable Engineering practice? Yes this is cool technology, but let’s stop selling hype and be transparent with all the challenges learned.
Can't wait to have a chauffeur ready to drive me anywhere I wish. It'll be so cool! I did have the opportunity to test out Tesla lane keeping etc and it's very impressive; albeit a little nerve racking... I see many videos coming out of America with these "self driving betas" - They are incredibly impressive! Such an insanely difficult problem to solve; I'm so proud of humans taking it on. Legends!
@@heyguysimcharlie
* Doesn't go from my house
* Doesn't go to my destination
* Crowded
* Cameras everywhere
* Often no seats
On the other hand... not sure cities would support that many uber taxis
@@tomwright9904 Walk to the bus stop 😂
Tesla admitted they are level 2 and will stay there with the current hardware. If you know the limits of a level 2 system, it can be a quite relaxing and save experience for everybody involved. Even if i wouldn’t buy a tesla (in fact i didn’t, but the brands i chose offer both great level 2 systems)
@@natenae8635 Walk? What is this odd word you speak of? We don't have that in America.
@@zaspalia I see must be one of those country exclusive products. Hopefully the bring to you soon.
Tesla: Some of you may die, but it's a sacrifice I am willing to make.
Do you know how many people die in cars driven by humans every day?
@@_Meng_Lan This is incorrect.
We should not have self driving cars on regular roads. Get specialized lanes with embedded guidance on long stretch highways. That's where it really makes sense.
Nonsense. Who would build infrastructure like that? Practically nobody.
The Mustang Mach E has BLUE Cruise..It monitors your eyes even thru sunglasses..If you are not paying attention it warns you...Best EV on the market..
It’s important because as people age, they don’t want to quit driving themselves to the doctor or the grocery store or even through the country on a Sunday afternoon. I hope someday I can get in my car put an address and start the car. I don’t care if I have to back it up all the time but when I get into the lane, especially if there’s no cars there I wanted to take off and drive follow the GPS what’s so hard about that
2:17 "with the exception of tesla, most cars need multiple sensors".
Who wrote this script? Teslas aren't powered by magic, they have 8 cameras and forward facing radar. They process this info with a computer the same way all self driving technology. The difference is tesla don't use lidar and the use of solely cameras and forward facing radar means there isn't a big sensor suite on the roof.
That's exactly what they said, you didn't understand correctly.
@@CanadianPrepper no they didn't listen to it again. They said with the exception of tesla's self driving needs sensors to pass data to the computers. Implying that they don't need sensors or even have computers.
I read recently that Tesla are no longer using radar either. Correct me if I am wrong.
I think it's misleading, but they meant that teslas can do self driving without any additional sensors equipped
@@ster2600 I would agree with that it is misleading. I think they know the right thing. To me tho, it should like your tell the public that tesla cars have nothing extra than a normal car and just work. But other companies they need all this extra stuff..
Elon has said it many times:
NEXT Year.
Never said next after which…
Wouldn’t it help if you could just disengage self driving in certain instances. Sometimes with just a command
That's how it works in a tesla you just grab the wheel.
I remember back then people were saying self driving cars would decimate the auto insurance industry as accidents would be almost eliminate.
It has. Tesla has 40 accidental gas pedal incidents a day. Each one stopping possible fatal accidents or just mabey bumper dents..
@@GWAForUTBE I’m sure it’s helpful and I know some insurance companies give a modest discount if you have that feature but it has not decimated insurance prices. One issue is that these features tend to be on expensive cars so the small amount of accidents are eaten by the high price tag of the car when the car needs to body repair or in total losses.
Besides the technical and legal aspects, the human one is often neglected. Anyone who owns a car wants to drive it. If not, you might as well take a taxi or public transport. The very essence of driving a car, as on a bicycle, a motorbike, a paraglider, etc. means being in control, being responsible for your vehicle, as well as for your life. This freedom is a powerful driving force in a society increasingly controlled by robots and computers.
"Anyone who owns a car wants to drive it"
Says who? A lot of people would love to have the option for self driving. A lot of people don't care about your definition of freedom.
Well actually when you’re driving a car you’re not in as much control as you think. You may be driving 100% safely but another driver around you might not have slept enough or might be drinking under the influence of alcohol or drugs or might simply be in attentive and driving through that stop light or stop sign and then unfortunately a crash
In the statistics in 2019 in United States if I remember rightly over 35,000 people died in a car crash because of the reasons as above. Some or most of these could be eliminated if people always drives safely but they do not but if a robot taxi was more reliable it would not be drinking alcohol or needing to sleep etc.
The first self-driving vehicles we see will be lorries, not cars. In a lorry it is no problem if the sensor equipment coast 20000€. And in difference to Tesla they do not work with cameras and artificial intelligence try to identify thing on pictures. They use sensors and if the street is not free they simply stop.
The UK needs its own "EV giant", not Tesla
hillarious
😂😂😂
Nothing can reach the best intelligence ever created. However, we can always try to come closer
this is like AUTO-AIM in a video game. there is no reason why it shouldn't work. Only obstacle IMO is dust/snow/rain etc. But beyond that it should work quite well.
Great, a car that auto-aims at the pedestrian crossing the road.
Videogame physics don’t have to deal with whatever edge cases might exist in real life, nor are the consequences for messing up as dramatic
@@alexsiemers7898 but is the final edge case better than the current system where millions die in auto accidents ? If we can eliminate even half that’s a huge win
We do have self-driving cars. You just have to be willing to ignore casualties if you use it. What incredible societal change.
Faded lane markings
Self driving cars doesn't solve cars, I'd much rather have more walkable cities and better public transport and just regular cars for enjoyment and edge case transport. And I'm a car guy from Sweden (northern Europe), not really the person to normally advocate for public transport
Because they don't press the accelerator when releasing the clutch.
Lidar and radar have to be bare minimum, even though its expensive
Self driving cars are the equivalent of making a robot change betamax cassettes over in a video player.
I only trust myself with driving not computer's there are so many things that can/will go wrong, just look at the electric cars they can combust spotane or when they upload energy ,and its a deathtrap when its in the water . Nah leave the driving to the people, but people really need to learn how to drive and know the rules of traffic. If you don't know these things.....please don't drive thats saver for everyone
Ai really develops faster than anticipated but we still have time before image recognition can go perfect and fast. Most ai developments happen in 6 months to 1 year nowadays so I think it will at least take 3-4 progressed algorithms to perfect the model which will take 2-4 years. And after that, they still need to implement the car so we have a long way to go.
Who is this "we"?
@@jimcrelm9478 Figuritive speech,"We" is for whole humanity.
@@exosproudmamabear558 Yeah about that...
It needs some more time, indeed. But it doesn't need to be perfect. It just needs to be better than humans, then the state will see a need to give way (in the true sense of the word) for the new technology (like the seatbellt). 2 - 4 years isn't long... give it another 5 - 8 years and we might have a different situation on the streets.
Self driving cars has to make life or death choices.
It’s time to put those misleading advertisements pay bigly about fully auto self- driving.
Elon in 2050: I am 1000% percent sure we will have self driving cars that can drive themselves from Newyork to London next year.
People don’t actually realise that AVs need connected technologies - Connected and automated vehicles (C/AVs). AV driving (and driving in general) can be safer by ensuring vehicles have the adequate supporting infrastructure for vehicle systems to have a real word version of the truth in terms of its environment, and what is happening on the road. Furthermore C/AVs very much depend on cellular infra (safety means low latency required), which as we know is hitting a blocker with 5G + the fact that before mass roll out you would need to make sure nearly all UK roads have excellent coverage. You cannot talk about automated without connected.
The people had enough once all other alternatives than diesel and oil had been explored and reached the conclusion that these are currently the best because they are the safest. The discovery that oil sources were being hidden as a result of The Antarctic Treaty left many customers angry. These car sales customers decided to boycott the electric and self - driving cars. And it was not just because of the treaty. It was also because of how easy it is to hack such a car.
I get frustrated because it is a complex topic and many consumers have been intentionally mislead by sources.
Proper AI vision systems will slowly continue to get better and better. Yes they might make dumb mistakes from time to time, but as more and more tweaks and updates are made it just continues to get better. Places like India that have unpredictable or unmarked road systems is actually where this technology will work the best and save the most lives because eventually the technology reaches a point it is better at threat detection and object avoidance than any person possibly can since no human can be continually looking 8 directions at once.
People will continue to trash on how the systems arn't safe enough. But then one day in the near future they will be. But they won't stop there. They will just continue to get safer and safer. That progress never truly ends and at some point these systems will become super human in capability, nearly never crash, and human drivers at that point may even be deemed a massive safety risk in comparison and outlawed.
It’s a novel idea but impractical in reality
This video just keeps talking about the once that are just not there yet and avoids talking about waymo aka level 4 self-driving level and only talks about Tasla
Since 2016 there have only been 31 crashes with automated drivers as opposed to how many crashes with regular drivers I will take automated cars any day. Automated cars do not have to be perfect they just have to be better than people.
Only 31 crashes, 19 fatal. With less than 1% of 1% of the global car market. If we scaled up global crashes to the 61% fatality rate of Tesla beta testers, you might see the problem.
@@LWylie And of those fatal crashes, how many were overriding the automated system warnings?
Ngl I do want a self driving car, but the problem is even if self driving has improved and any of us use it, we're still surrounded by other drivers which it might be risky.
It needs to be one way or the other I think
@@mikeh2006
if they really want this to work I think they should make these cars able to communicate with each other, even if new driving cars if possible. If it's not possible then they should banned driving cas or self drivng cars just won't happen.
This is why Apple uses the iPhone as a platform for sensor development. Lidar on mobile perfect machine vision, real world scale test it on billions of devices. Adding a battery and a motor is the easier to do when the software stack is so stable. Millions of Devs unfortunately I think Tesla needs that super computer.
Thats not how the real world works.
maybe the solution is to make an autonomous highway too...
your forgot to mention mobileye
Didn't Google use self-driving cars years ago when they we making StreetView??
They planned to, but never got around to merging to two. They went into high-res for the mobility group, which is why they spun it off into Waymo.