Best summation on why Self Driving cars are a long way off

I'm sure this will trigger no one. The popcorn is in the microwave as we speak.

I agree that usable, fully autonomous cars are not here, now.

But it's hard to predict if or when they will be. On one hand, we went from Orville and Wilbur Wright's first powered flight to a manned moon landing in just 66 years. On the other hand, the 1950's dream of practical, Jetsons-style flying cars remain a dream, roughly the same number of years later. (Please, don't use Terrafugia as a meaningful example!)

I was a teen computer geek in the 1970s, and at that time, nobody dreamt of a time when:

...it would be common for many North American individuals to own a computer that's more power than all of NASA's computing capability

...that computer storage would be measured in terabytes and memory in gigabytes

...that this computer would be interconnected (networked) with the rest of the world and have access to massive volumes of information

...that you could "stream" high-definition video on music, on demand

...that this computer would be about the size of pack of cigarettes, and less than half as thin, and also be a mobile phone.

Yeah, it's pretty hard to predict the future, but we do know that today's autonomous cars are certainly not ready for prime time.

Typical faulty apples-to-oranges illogical reasoning.

I feel the biggest problem with Self-driving cars will be Insurance. WE already see that insurance for young new drivers can be many times that of long term safe drivers - largely because of lack of history.

Now - we have self driving cars - with which we have no real history. But - the problem is - there are lots of other considerations to the car -and the conditions - that will have to be considered as well. Example -= would driving in a Thunder/lightning storm be a extra problem? Certainly - condition of - and maintenance of a vehicle will also come into play as well. And I assume as these things age - will older less sophisticated systems become more expensive as they are replaced with later generation systems.

This MAY be a situation where the car would need insurance as well as the Driver. I would expect this to cost more money than the average person would afford early on in the cycle - so it will take a long time for the technology to produce an Affordable situation.

@TomM That and the cost of litigation in the plethora of lawsuits that will happen.

TomM wrote "...WE already see that insurance for young new drivers can be many times that of long term safe drivers - largely because of lack of history...."

My understanding is that young male drivers have high auto insurance rates because history shows us they are famore more likely to end up in an accident, and thereby make a claim on the policy. So the have lack of driving experience (an individual's driving "history"), but the demographic (young male) have a history and reputation for being accident prone.

They forgot simple environment problems that will be extremely difficult to decide. Snow/sleet/ice combined with road salt reside on the roads block the vision and sensors of the systems.There is a reason my auto cruise control gives me the message " I can't see;I have to shut off" when my car is on the highway and the sensor has an inch of icy slush on it. Lane lines? What lines as nobody can see them and the road becomes follow the first guys tracks.

Matt, please pass the popcorn. Thanks.

To hell with popcorn, let's have martinis.

I don't think anyone ever said creating self-driving cars would be easy. Just because it is a challenge doesn't mean we shouldn't do it. Waymo effectively has a level 4 car right now, Tesla AP2 is already great on the freeway and they are starting to push features for off-freeway roads (which admittedly is where the real challenges are). I think long way is more like a couple to a few years, it will comes sooner than you think.

It will be regional city by city dependent on environment and traffic density and layout. Bay Area, LA, Phoenix etc first. It will take much much longer in places like Detroit (French hub spoke roads with and English Grid on top, plus winter) or Boston (paved cow paths, 5 and 6 road intersections, wild drivers, winters) and places like that.

It is the only car you can get with true lane keep below $100k. I hope you can realize the difference between a beep when you are going over the line and the car actually steering for you.

Self driving cars will REALLY be a long way off once automakers (cough Tesla) start rolling out more autonomous features before they are properly tested and the body count starts climbing upwards.

So far, most "autonomous" victims are douches driving EV's while watching "Barney" videos. Once innocent bystanders start getting mowed-down, watch out!

This battle will not be lost in the technology lab, it will be lost, prematurely, in the court of public opinion.

You nailed it.

Wake me up in 10+ years when there is a full solution. In the meantime what is the point of trying to relax as a passenger - occasionally to be suddenly prodded in the last second to save a wreck?

Reminds me a lot of driving with a student driver. If you're not feeling a good measure of stress as the passenger, you're pretty naive.

People are too busy trying to find flaws and reasons not to do autonomous driving, but forget one main reason FOR doing it: the high cost of HUMAN driving - 100 lives lost every day in America alone, over 1 Million worldwide annually.

Progress always takes time and commitment. Regardless of what is said here, progress is already taking place. Major players are already committed and regulators are reluctantly joining force. You can either be a part of progress now, or wait to get on board later when everything is in place.

With AI growing so fast and so advanced now that Level 5 autonomy is not too far away - I would say 5 years top for the major players such as Tesla, Waymo, Apple, ...

Another part of the argument is that "we" human drivers consider ourselves "the most complex system on the vehicle" but forget that "we" consists of a bunch of drivers with varying level of driving skills, situational awareness, mental stability (BobM and MDarring are automatically disqualified), ...

Consider an ideal driver - one who maintains full control of the vehicle at all times, who has perfect awareness of the surrounding environment, who can make decisions in milliseconds, who can "see" traffic several cars ahead while being blocked by larger car in front, and who always remain calm, ...

Unfortunately such human driver does not exist. The machine driver may not have the capability of the best driver at the moment, but it can learn and is learning really fast.

The human, on the other hand, does not seem to learn much after getting his driver's license. Depending on who you talk to, that person's mental state can be the worst part of the "most complex system on the vehicle" and can get people killed on the road.

I absolutely agree with you that we engage in fallacious thinking: '"we' human drivers consider ourselves 'the most complex system on the vehicle'".

Driver error is a massive contributor to mishaps.

@HenryN I don't know if you've done performance driving classes as I have, but even though I am pretty competent on the track, I've still been bitten by driver error.

Ironically, track driving is a cake walk compared to street driving and street driver have far less training.

That said, I do not see the solution as autonomous cars.

I do see a tremendous benefit in using some of that automation to better assist the driver.

I think there is a tremendous benefit to vehicle telemetry where the vehicles in a given circumstance are communicating with each other.

Self-driving cars depend on artificial intelligence capable of entirely duplicating what humans do. The leaders of advanced artificial intelligence institutes are on record as saying we are two decades away from that capability. One leading institute found that when trying to develop a system that could pass a standardized test elementary students take they could get to a 70% performance level. The other 30% depended on subtleties in human experience and judgment that at this point cannot be duplicated. So for those who say it will happen quickly, how will they solve that problem. Self driving cars that operate at the level of a 5 year old human?

Hey Bob, if you have to bring up "my 40 years as an Design engineer, I can promise you this...", you already lost the argument and grasp for straw. You are still at the first 2 stages of the disruption denial - Confusion and Repudiation. Your tendency to cite obscure articles doesn't help you either. The article you post is nothing more than basic knowledge of self-driving, nothing ground breaking there. The guy sells simulation software, and he is promoting software simulation for self-driving - what a surprise!

In case you're not up to date, Apple is finishing up another "Spaceship" building in Sunnyvale CA which will house their automotive research. Waymo is Alphabet's brand new subsidiary focusing on autonomous car development. Tesla already has over 2 Billion Miles of Autopilot data (HiL), and the number is growing even faster with the new Model 3. Adding Nvidia and Qualcomm on the hardware side, the AD landscape is a lot more interesting than you can imagine.

Silicon Valley is now abuzz with tech companies large and small working on AI and autonomous driving (even the ultra conservative Ford and GM spent $1B each to acquire startups Argo and Cruise to jump start their self driving effort).

If you have to bring up technology references from 40 years ago, you're apparently out of date.

@Bob - MD said this article would "trigger no one". Guess he was wrong.

What triggered your "web rage" ? Someone disagreeing with you ? Mention of Tesla ? You out of date ? or simply you forgetting to take your meds this morning ?

It's tough to be old and be filled with rage Bob. Time to take some time off and go see Silicon Valley to open your eye ?

@BobM: check the news today on the alliance between Apple-Hertz and Waymo-Avis, and close your pipe you old fart. The more you talk about your "knowledge", the more it shows you're both a "never-has-been" and a "never-will-be". I will give you some slack if you sit and keep quiet.

The term "take your meds" is an insult to normal people, but for you Bob, please Take Your Meds - seriously! You are getting senile.

You screamed for "REAL F'N WORLD" data - but that's exactly what the likes of Tesla and Google have been doing for years. Tesla has amassed over 2B miles of data with AP hardware in the loop - the majority of which is in "simulation" or "shadow" mode - just as you asked. Google and Apple are also doing or trying to do the same thing, albeit at a much smaller scale.

You proudly presented the article as if it were something ground breaking. News for you Bob, it's nothing new there. If the article was written 10 years ago, then it may deserve some merit, but June 2017 ? Try harder next time Bob.

The cost of travelling 4.3 Trillion miles is 35000 lives - per year. How many people died using a flawed Tesla Autopilot (AP1) ? ONE over the course of 3+ years - and that fatal crashed was confirmed a blatant human error.

And about your math, you off by only ONE significant digit. I assume, as a "Design Engineer", you should know what that means.

Just curious, for an opinionated guy like you, why would you care about up or down arrows ? if it makes you feel any better, I'll give you an UP arrow.

+1 for BobM

Few things excite me more than a series of tight curves flowing in and out of a beautiful piece of open road, but the writing is on the wall.

There are too many parties with a vested interest in getting MORE autonomous on the road sooner than later. Parties with more money and political influence than anyone person or group on this board.

Infants today will probably be the last generation to learn how to drive.

People don't want to think.
They want to live in mediocrity.
They don't want to take responsibility for their actions.
They are lazy physically and intellectually.
They love autonomous cars.

@md

Separate issues.

For a majority of drivers, their time behind the wheel is spent in traffic, a situation that will only get worse. Add into that no open roads, increasing costs of ownership, an ever shrinking space to park/store a car and both the proliferation of home delivery and "work from home" jobs/shrinking retail driving has become an inconvenience to a growing number of people around the world.

It's not about being lazy, or any of the other things you listed.

Society is simply evolving/moving/transitioning beyond personal car ownership.

In 20 years, the only people that will be driving themselves will be the rich and the poor.

. . . shrinking retail jobs will only reinforce to many that driving has become . . .[error due to lack of coffee]

Side note, there is a great book about AI called Manna that tells two stories about how AI could evolve, one is vastly negative and the other is positive. Should check it out.

Autonomous cars don't need as much "AI" as people believe. Level 4 doesn't have to be perfect, it just has to be at least 10 times better than a human.