Articles, Blog

Why Are Self-Driving Cars Taking So Long?

Why Are Self-Driving Cars Taking So Long?

SciShow is supported by By now, you’ve probably heard that self-driving
cars are coming soon. If you haven’t—surprise! They’re coming soon! But people have been saying that for at least
a decade, and I still can’t buy a car that’ll drive
me to work while I nap in the passenger seat. Some cars already come with partial autonomy, systems like Tesla’s Autopilot, that assist drivers or sometimes even take
control. But they still need a human driver who can
grab the reins on short notice if things get dicey, which is why someone
in the UK got arrested earlier this year for trying the passenger
seat thing. There are some fully driverless vehicles that
might be released in the next few years, but they’re only
meant for very specific uses, like long-haul trucking or taxis confined
to certain streets and neighborhoods. That’s because general-purpose driving is
hard! The software has to work out a lot of really
tricky questions to turn information from its sensors into
commands to the steering and pedals. And despite all the money and brainpower that’s
being poured into research, there are still major challenges at every
step along that path. The first thing a self-driving car has to
do is figure out what’s around it, and where everything is. It’s called the perception stage. Humans can do this at a glance, but a car
needs a whole cornucopia of sensor data: cameras,
radar, ultrasonic sensors, and lidar, which is basically detailed 3D
radar that uses lasers instead of radio. Today’s autonomous vehicles do pretty well
at interpreting all that data to get a 3D digital model of their surroundings the lanes, cars, traffic lights, and so on. But it’s not always easy to figure out what’s
what. For example, if lots of objects are close
together say, in a big crowd of people it’s hard for the software to separate them. So to work properly in pedestrian-packed areas
like major cities, the car might have to consider not just the
current image but the past few milliseconds of context,
too. That way, it can group a smaller blob of points
moving together into a distinct pedestrian about to step into
the street. Also, some things are just inherently hard
for computers to identify: a drifting plastic bag looks just as solid
to the sensors as a heavier, and more dangerous, bag full of trash. That particular mix-up would just lead to
unnecessary braking, but mistaken identities can be fatal: in a deadly Tesla crash in 2016, the Autopilot cameras mistook the side of
a truck for washed-out sky. You also need to make sure the system is dependable, even if there are surprises. If a camera goes haywire, for example, the car has to be able to fall back on overlapping
sources of information. It also needs enough experience to learn about
dead skunks, conference bikes, backhoes sliding off trucks, and all the other weird situations that might
show up on the road. Academics often resort to running simulations
in Grand Theft Auto yes, that Grand Theft Auto. Some companies have more sophisticated simulators, but even those are limited by the designers’
imaginations. So there are still some cases where perception
is tricky. The really stubborn problems, though, come
with the next stage: prediction. It’s not enough to know where the pedestrians
and other drivers are right now the car has to predict where they’re going
next before it can move on to stage 3: planning
its own moves. Sometimes prediction is straightforward: a car’s right blinker suggests it’s about
to merge right. That’s where planning is easy. But sometimes computers just don’t get their
human overlords. Say an oncoming car slows down and flashes
its lights as you wait for a left. It’s probably safe to turn, but that’s
a subtle thing for a computer to realize. What makes prediction really complicated,
though, is that the safety of the turn isn’t something
you just recognize it’s a negotiation. If you edge forward like you’re about to
make the left, the other driver will react. So there’s this feedback loop between prediction
and planning. In fact, researchers have found that when
you’re merging onto the highway, if you don’t rely on other people to react
to you, you might never be able to proceed safely. So if a self-driving car isn’t assertive
enough, it can get stuck: all actions seem too unsafe, and you have yourself what researchers call
the “freezing robot problem.” Which itself can be unsafe! There are two main ways programmers try to
work around all this. One option is to have the car think of everyone
else’s actions as dependent on its own. But that can lead to overly aggressive behavior,
which is also dangerous. People who drive that way are the ones who
end up swerving all over the highway trying to weave between
the cars. Don’t do that, by the way. Another option is to have the car predict
everyone’s actions collectively, treating itself as just one more car interacting
like all the rest, and then do whatever fits the situation best. The problem with that approach is that you
have to oversimplify things to decide quickly. Finding a better solution to prediction and
planning is one of the biggest unsolved problems in
autonomous driving. So between identifying what’s around them, interpreting what other drivers will do, and
figuring out how to respond, there are a lot of scenarios self-driving
cars aren’t totally prepared for yet. That doesn’t mean driverless cars won’t
hit some roads soon. There are plenty of more straightforward situations where you just don’t encounter these types
of problems. But as for self-driving cars that can go anywhere… let’s just say the engineers won’t be
out of a job any time soon. I love the layers of thinking involved in
this kind of problem solving. And while I’m not an engineer designing
self-driving cars, but I still get to practice this kind of thinking
on Right now, I’m working through the Convolutional
Neural Networks lesson to help me learn how to work with neural networks. I’ve already gone through the overview, and this “Applications and Performance”
quiz has a car on it, so that’s what I’m going to try my hand
at next. The quiz already explained how this network
works. And then it’s asking how we should modify
it to suit this imagenet challenge, to help it categorize objects better. I think the answer is C: to add a fully connected
network at the end to help predict probabilities for what the
object is, based on the high level filter activations. And I got it right! What’s great about these quizzes is that
they keep building on each other, so even though I got that one right, I’m still getting more information throughout and each question gets a little bit more interesting. And if you get one wrong, that’s ok too! Because the point isn’t to beat the quiz,
it’s to keep learning, just like these neural networks do! So, if you want to test out YOUR neural network, the first 200 viewers to sign up at
will get 20% off their annual premium subscription, and you’ll
help support SciShow – so thanks!

Tagged , , , , , , , , , , , , , , , , , , , , , , ,

100 thoughts on “Why Are Self-Driving Cars Taking So Long?

  1. The first 200 viewers to sign up at will get 20% off their annual premium subscription and be supporting SciShow!

  2. My understanding (based on an article I read years ago – can't remember the source) is that the biggest impediment to self-driving cars being successful is cars with drivers. As long as the vast majority of cars on the road have drivers behind them, self-driving cars can't be successful. As this video points out, the human factor is too hard to predict. We need to get to a place where the vast majority of cars are self-driving before self-driving can really take root.

  3. All the evil corporations are willing to sacrifice a few people in order to get there first. Huge profits await.

  4. They had better figure out the fully-autonomous driving thing soon, because we are soon headed for situations in which no human can take over if the computer becomes befuddled.

    • "I am just a child. I do not know how to drive."
    • What human? There is nobody in the car.
    • "I am asleep. Do not bother me unless it is something important."
    • "I am too old to drive. I can't see well enough and my reaction time is too slow."
    • "I haven't driven in years, and I hardly remember how."
    • "I have no driver's license, nor any experience driving."

    Those scenarios are probably coming very soon. Probably many old people who can no longer drive but want their independence, will be buying self-drive cars. Also, due to self-drive capability, car-sharing could suddenly become a huge thing if it turns out to be cheaper and easier than car ownership. Families will also be able to share cars among all family members and with their friends. After running an errand or taking somebody to work, the car can return home empty or run the next person's errand in the meantime.

    One possibility is that self-drive cars can call up the "mother ship" service in the internet cloud, to help it navigate a situation that it just does not understand, rather than just stop in the middle of the road and wait hours for somebody to come and intervene. A human could tell it what to do, or drive it a short distance, by remote control. However, that is an experimental phase stop-gap, that probably self-drive cars should not need to rely upon much as they get more advanced in the future.

    I think that ultimately, computers will drive far more reliably and safely than humans do, having some huge advantages of powerful computing power, the ability to not become drowsy, and being able to remain consistently alert to all hazards for many hours of non-stop driving, something that no human can do very well. After many situations of not hitting the dear, of skillful collision avoidance maneuvers, more and more people will start saying things like, "I will never again own a car that is not self-drive." Although I am a very good driver, I am tired of driving, and I would love to have a car that is quite capable of driving itself. I could enjoy so much more beautiful scenery than I could with my "eyes glued to the road". Driving is such a needless boring chore, much better suited to computers to do. I feel like my brain is running a computer program when I am driving. Sorry, but the human brain was not designed to run computer programs. It is a serious mental load that causes a lot of stress, especially in bad weather. I would feel so much better to be in a reliable self-drive car, that can easily see through fog and glare and rainstorms and blizzards. And as the network of self-drive cars grows, self-drive cars will be constantly communicating with one another, pre-alerting other self-drive cars to all the potential hazards out there. Just as soon as one wheel starts to slip a little on icing roads, all the self-drive cars in the area will instantly know about it, and take that into consideration.

    Self-drive cars will greatly boost the efficiency of already-existing roads and avoid traffic congestion bottlenecks.

  5. I think to solve this problem a Universal Car Software should be made and it should be installed in every new car. This software will allow it to communicate with other cars having this software being able to send it's state and the actions it is going to take to every other Car in its vicinity. This can be implemented with Fly/Drive By Wire Technology which would allow computer to know Hunan's actions beforehand. This type of technology has been seen in video games where AI objects move randomly but don't collide with each other.
    So this way all people don't have to buy Self-Driving Cars. Normal cars can communicate with self driving cars.

  6. You could give autonomous vehicles the simplest A.I. and I'm sure it will perform roughly as well as most drivers here in Manitoba. All of that extra stuff goes way beyond what the majority of drivers here are capable of 😉

  7. The world needs standardised road rules and road conditions for self-driving cars to be easier to be made faster

  8. I don't want a damn self-driving car. This whole society is trying to take away my money, my rights, full control of my life and my self worth. and by God I'M KEEPING CONTROL OF MY CAR.

  9. Well, Waymo have shown completely autonomous cars for 2 years now. I don't get your point. 🤔

    You don't want to buy a car from Waymo, they are too expensive, but you can participate in the early access program, which now runs for nearly a year and just call a car, if you need it.

  10. As long as AI-driven cars result in a lower percentage of accidents than human-driven cars (which they do), then AI drivers are superior to human drivers.

  11. Self driving cars take the defensive driving approach, that's why they can drive around other people just fine.

    Also, why do we assume that a car with 8 cameras, multiple ultrasonic sensors, forward radar that can see through bad weather and a computer that doesn't get tired or drunk can't drive as well as a sack of meat with one or two eyes and a couple of mirrors?

  12. I think open simulators like or deepdrive could be an interesting approach to bring a larger piece of the academia on thinking about interesting autonomous driving situations.

  13. They should probably figure out GPS first! I mean the Uber driver app tells me to drive through buildings, fences, and rivers thankfully i am smart enough to go around even though the Uber app say go back and drive thru the building, fence or river! These autonomous vehicles have sensors that wont allow it but if the GPS says to go thru will the car just sit there???

  14. I'm far from being a Luddite but I can't see truly self driving cars ever happening. It's in the news today that the driver in the fatal Uber crash was watching TV at the time of the crash and could be facing a manslaughter charge. Ultimately what's the point of a self driving car if you can't watch TV?

    There's two possibilities for self driving cars… Either the person in the driving seat is legally required to devote their full attention to the road and be prepared to take over immediately (which means hands on the wheel and foot hovering over the brake pedal) or the car itself is legally responsible. In the first case why not just drive yourself and in the second case how do you hold a machine legally responsible for its actions?

    I don't buy the argument that statistically speaking self driving cars are safer… I think it's based of faulty logic. One self driving car may be safer than the average of all cars on the road but would that still be the case if, for the sake of argument, 50% of the cars were self driving? I'm guessing not.

    If we were starting from scratch and all cars were self driving and all followed the same tightly defined set of rules then maybe things would work out fine but going from all human driven to all robot driven involves going through a long period of there being a mix… And that's where the problems start.

    If self driving vehicles are such a great idea why aren't all trains driverless by now? I'd say the answer is that a human is needed for situations the computer can't cope with… And there goes the justification for having robot drivers.

    PS: Yes… I'm aware there are self driving trains but they are rare and usually part of a system built from the ground up to work that way.

  15. Self driving cars are being held back by the reliance on Lidar. Lidar is junk. It is not to be trusted. So many companies are trusting it that self driving is doomed to fail. Thank God for Elon! The only engineer that understands that Lidar is broken and relying on it is foolhardy. Too many "businessmen" think Lidar is similar to Radar just because they sound the same. They are not. Lidar is a laser pointer flicking on and off trying to see a picture of what is in front of it. Radar is a wave that bounces back a full image. HUGE difference! Lidar is trash. People have died and will continue to die because businessmen are making decisions about things they do not understand. Buy Tesla or nothing.

  16. First of your videos I've watched. I use a Sony Bluray that has limited smart functionality and no ability to comment. I want to buy a Level 3 autonomous truck.

    Excellent video, your speech is crisp, well researched content and decidedly superior to 90% of the available YouTube videos. My first sort is the video channel. I'll be looking for more. Thanks

  17. I loved how you simplified such a complicated matter. 🙂
    I love Engineering and that's why I am studying it.

  18. What about the 1/2 of the world that won’t except this what next? A Rebellion like Star Wars that would be cool… well if I can control the speed we good 😎😎I’ll just take off the governor that you know will be on it😏

  19. I hate self driving cars like they say “the more crutches you have more it hurts when kicked from under you”

  20. I personally believe that self driving cars are not necessarily needed. Humans have been driving manual driven cars for almost a thousand years if not more. Why change something that is already good enough. Too me it’s just another way for us Americans to be lazy compared to a lot of other countries.

  21. People expect perfection out of self-driving cars. It won't become mainstream until self-driving cars reach that perfection.

  22. In general this video is pretty good, however, you guys didn't mention the fact that Google already figured out self driving cars.

  23. If you take humans out of the equation there would be much less problems. The transition over to full autonomous driving on the roads will take decades until humans wont be allowed to drive. It will never be perfect but it will be eventually much safer. less human error.

  24. Cool!    I hope to ride in one soon .   Patents 672256 3190554 , 3013505

  25. Machine learning still equals machine programming. Humans are born with a level of common sense to identify if something doesn't look right even if we don't yet know what it is. Dynamic problem solving is something we do every day without thinking about it. Like this guy said, humans immediately recognize all the different objects and can quickly make the correct decision whereas a computer will have gather that data, ask its catalog what they are and depending on its programming will commit an action (could be wrong or right). The computer doesn't know wrong or right even they only know what their programming tells them regardless of how sophisticated that program is. Saying the phrase "Artificial Intelligence" is just saying a computer has large database of factors, equations, and solutions.

  26. Automation is way of and I'm very glad after hearing about those bad accidents. Needs a lot of refinement before I'll trust it. I can understand completely why people want them though. I was on a very long motorway journey on my motorbike not long ago and the roads were so quiet too so I was weaving in and out of different lanes to keep myself from falling asleep, it's dangerous to be that bored 😒

  27. To me, the biggest problem with self-driving cars is its inability to make complex decisions. Driving requires a lot of decision-making skills that computer don’t have. Why do you think you have to take a test to get your driver’s lisence?

  28. Your not supposed to say never but here goes . . . . There will never be driverless cars. NEVER. Because this is a tough solution for no problem. If you ask anyone to name their top 100 problems, driving one’s car would not make the list. We can split the atom, we have been to the moon, but why invest in this giant undertaking that will never be as good as what I can teach a 16 year old in a weekend.

  29. Lets see how human laziness has progressed: 1st, we make cars to get us places faster…fair enough. Next, we have cabs cause we don't feel like driving the cars…finally, we make cars that drive themselves cause we don't want to do it LMAO

  30. I know just barely enough about the technology and science of all this to understand that all those things he mentioned are, in fact, very extremely awfully difficult, and every hurdle we pass in the goal for autonomous cars is an incredible leap.

  31. Self-driving cars are going nowhere….& some might find it interesting & informative to view just a couple of the many Youtube videos on this subject…namely…"The driverless car scam"…&…."The terrifying truth about self-driving cars"

  32. Maybe in 100years this selfdriving era wll completely take over the planet,untill than its just an good idea that needs lot of improvements and changes..people who are now in their 30s wll not feel the benefit in their lifespam..

  33. I don't get it. Why don't we just build or restructure current roads with sensors underneath them designated for self driving cars ONLY. Is it just too big of a switch to be feasible? Maybe once city switches or they start with one highway across town and go from there.

  34. I think a mass transit would be more effective. I didn't like cars to much before, but Adam Ruins Everything showed a lot of flaws in everyone needing cars

  35. I think Programmers are over complicating things. If the car is at a 4 way intersection and doesn't know what to do. The computer should just signal to the other car to get out of its way. And there should probably be some type of light on the car to indicate its self driving so drivers can be aware.

  36. I see two problems with self-driving cars: 1) If it comes down to a choice of hitting an old lady jaywalking vs a bus full of children, in a situation where another self driving car or regular car prevents it from braking how does it make the decision and what about insurance? 2) Imaging getting into a self driving cab where a group just had an orgy or being the person who has to clean up the mess?

  37. All weather self driving cars wont be available anytime soon until developers come up with an all weather sensor. Right now these cars can only drive on clear days.

  38. The military is using these frequencies at 95GHz as croud control tech, and the 5G networks will operate in the range of 24-90 GHz you should inform your self on the health dangers and show this to Elon or we will be put inside a giant microwave. These frequencies put the heart and brain out of sync with the Earth's natural schumann resonance, which is our biological network of operation. I mean people should be informed

  39. Or we can make the car ask like “overtake car ahead?” And you press a button to do it or like “switch lanes?” “Turn now?” Like it’s still self driving and safer.

  40. Why are self driving cars taking so long? Well hopefully it’s because the people behind the development of the technology have been weighing the pros and cons and are realizing just what a horribly BAD idea it is.

  41. If only we could change everything to self driving car then the computers could just talk to each other and we wont be worried that the computer cant predict our human actions

  42. Here's a practical solution for everyone to love:
    Why not force the shitty drivers who value looking at their messages or drinking while driving or who love to speed and cut people off, to be in self-driving cars and have the enthusiasts and sane, safe, lawful drivers continue to drive their cars? That way we can all live together in harmony on the road.

  43. If you ask yourself this, who would get the driver's license? It would still be you, you know, in case of when the car glitches up or fails to turn, you would still have to have knowledge of the road to intervene in those situations. Legal framework would revolve around potential danger and hazards. Self-driving doesn't mean you don't have to learn how to drive and no have to put your hand on the wheel.
    The adoption for the masses will take decades. You think your average joe will fork tens of thousands more on a car that acts autonomous, when he can most likely drive well and sees no reason to get rid of his car?
    And let's get real, we still live in a world where people still drive stick shift, you think those drivers will get a car that stripes away all control for an expensive price tag?

  44. too many variables to have a safe self driving car ..things in suburbia come out of nowhere unpredictably…kids ..dogs ..balls ..birds …come out of nowhere and the car will not react to this good luck people its not a good idea

  45. The only way I'd feel safe being in a self-driving car is if the roads were smart too and ALL cars were tapped in to the computer grid sensors and stuff. But now were talking crazy expensive. It's cheaper for mass PUBLIC transit to be more widely available and more affordable. But that's just me.

  46. Only an imbecile would ever consider entrusting his one and only life to a non human "intelligence".
    No one has any interest in riding in a driver-less vehicle…period…the end.

  47. I don't understand why people are so hype about how driverless car will reduce death rate by car accident. All driverless car did was make sure that a drunk or an easily distracted person don't drive. You take those people out of the equation, you reduce car accident by 97%. Only 3% of car accidents are caused by healthy humans and attentive drivers. Do your homework

    It's not that driverless car doesn't make mistake. They give irresponsible people the option to let a machine drive for them. But considering the fact that those kinds of people is on the lower side of the economic ladder, chances are they can't afford a 30K driverless car.

    Instead of teaching people how to be more responsible, we are investing millions, if not billions, of dollars to invent driverless car. At the same time, we are in no way taking accountability for the accident that comes with testing those cars. People on the internet (especially reddit) are literally saying that "Those accidents are for the better goods."

    Not to mention the liability issue. If the company that sold you the car is "at fault" when the driverless car get into an accident. You can expect insurance prices affecting the actual retail prices of the car. If I give driverless cars all the benefits of the doubt, and assume all the following:

    – Company is "at fault" during accident, and so a commercial license is needed
    – The tech and softwares that gets add on to the vehicle at $5K max
    – Additional investment on a repair center specializing in those cars

    We can expect driverless cars to be at least 30K for a price tag. And cost of ownership to be slightly under the cost of ownership of a traditional vehicle. Financially, it makes ZERO sense to buy a driverless car. But that's never an obstacle since Americans have proved to willingly destroy their own livelihood for a car and tech. You would have to punch a temporary 57yrs (on average)hole in your finance to do all that.

  48. Self driving cars are coming but I think they'll start on the freeway. The freeway is for the most part straight (no major turns, just curves unless your entering and exiting). Driving on city roads is hard even for a lot of human drivers (how do some people get drivers licenses?) I think for starters vehicles should communicate with each other. This way every vehicles nows where the other is. This can be done in a few ways. If you have a self driving car each computer will just be able to see each other. If you don't have a self driving car you can either 1. get a dongle to hook up to your computer or 2. it can use a smart device that you have on you (phone, watch, tablet). Some people may not want this because they'll think it will hamper security, but it doesn't have to know much information, just that you're a blip on a map. If you're in a vehicle it'll know by speed and the way you move and where you are if and same if you're walking. GPS' now know if you're walking or driving. It doesn't need to know your name, the make of your car or any other information, just that you're a vehicle or pedestrian moving and where you are. The issue with this is with GPS as it is it won't be accurate enough so we have to wait for GPS 2.0 to come out, because it would be AGPS (and GPS) which is accurate by meters when GPS 2.0 will be accurate by centimeters. GPS' accuracy is one reason we don't have driverless cars now, the military may have GPS 2.0 now (not sure), it'll just take time until the general public gets it.

  49. Never mind self-driving cars, why are flying cars taking so long? We were supposed to be flying them about fifty years ago.

  50. I like how hundreds of thousands of people die every year in driving related accidents but this video wants automated cars to be absolute perfect in every possible scenario that could ever exist. We need to wrap our minds on the correct goal. Automated cars don't need to be 100% perfect in every possible way. They just need to be better than the average driver. There will still be accidents, you won't ever be able to predict everything, but if automated reduces them, we are already winning.

  51. For all who are asking that stupid question who is to blame if the autonomous car violates traffic laws or gets in a crash… for one, autonomous cars aren't going to be violating traffic laws so if one is, that is definitely not the AI but the reckless driver behind the steering wheels fault. Second, you are responsible for supervising the road and your surroundings regardless if the autonomous feature in the vehicle is active or not. Failing to do so and the car getting into an accident would be considered negligent supervision of the vehicle putting you at fault.

  52. I refuse to have a self-driving car until they make them capable of speeding AND swerving around potholes.

  53. does like half this comment section forget our current state of driving?? tens of thousands of deaths a year, and hundreds of thousands of injuries, yet you're concerned about a computer misjudging a plastic bag?? haha give me a break

  54. Self driving cars can detect their surroundings but that doesn't mean that they understand their surroundings. E.G. on a residential street a ball comes bouncing into the street with no chance of the car hitting it. Most anybody will break knowing a kid is right behind it. But unless this scenario is explicitly programmed in, the driverless car will keep going.
    Computer security will be essential to prevent hacked cars from being used in murders and even mass destruction.

  55. I gave people the idea for self driving cars. But in order for it to work the roads have to be equipped. For it to work a lot of the cars have to be self driving so they work together. Replacing the current manually driven vehicles is a important and most difficult part of making the new system work.

Leave a Reply

Your email address will not be published. Required fields are marked *