What's new
What's new

Self driving cars in the present, good idea or bad idea?

  • Thread starter Guest
  • Start date
  • Replies 175
  • Views 22,160

Self driving cars in the present, good idea or bad idea?

  • Good idea

    Votes: 10 35.7%
  • Bad idea

    Votes: 18 64.3%

  • Total voters
    28
G

Guest

Guest
On another forum where I am probably the only member who knows much about the systems contained in a self driving car there are a few people adamant about how wonderful they are and have no understanding of the possible problems no matter how they are explained to them. So I would like to ask a group of people who understand the possible disasters that lurk when components in motion control systems fail.

I am asking you consider your opinion reflecting the current state of our infrastructure and not a road system 20 years in the future designed around autonomous vehicles. My opinion is they are a bad idea, they finally killed someone, while the car would not be judged at fault a human driver probably would have avoided hitting the pedestrian, the autonomous car neither braked or swerved. They have a record of being involved in an accident every 50,000 miles, rarely judged at fault but these were accidents a human driver probably avoids. They are almost exclusively tested in California where conditions are mild compared to other states.

As all of us in here know this fleet of prototypes being tested are getting top notch care that the average autonomous car won't get if they become common place and mass produced. Also what will happen when they start to age? They have an accident every 50,000 miles when they are brand new and probably maintained by engineers and top level mechanics. What will happen when driven by Joe Six Pack who waits till his brakes make grinding noises and takes the car to Jiffy Lube?
 
IMO they are a very immature technology being pushed on the public for political reasons.

What too few of the fan boys realize is what they actually "see". A Tesla on autopilot drove straight into the side of a tractor trailer that would have been easily recognized by a human. Now a self-driver Uber vehicle ran over a woman more than 2/3 of the way across a street.

As usual, the boosters in the media put spin on it by saying she "came out of nowhere". That is a blatant lie as the video shows she was walking across, not running or riding, and a human driver would have noticed her in time to brake.

"Joe sixpack" will never own a SDV. These will clearly be fleet operated vehicles like cabs and a very wealthy person might be able to lease one full time with maintenance performed by the company but for ordinary people it will just be another form of taxi, albeit one that puts humans out of work.
 
On another forum where I am probably the only member who knows much about the systems contained in a self driving car there are a few people adamant about how wonderful they are and have no understanding of the possible problems no matter how they are explained to them. So I would like to ask a group of people who understand the possible disasters that lurk when components in motion control systems fail.

I am asking you consider your opinion reflecting the current state of our infrastructure and not a road system 20 years in the future designed around autonomous vehicles. My opinion is they are a bad idea, they finally killed someone, while the car would not be judged at fault a human driver probably would have avoided hitting the pedestrian, the autonomous car neither braked or swerved. They have a record of being involved in an accident every 50,000 miles, rarely judged at fault but these were accidents a human driver probably avoids. They are almost exclusively tested in California where conditions are mild compared to other states.

As all of us in here know this fleet of prototypes being tested are getting top notch care that the average autonomous car won't get if they become common place and mass produced. Also what will happen when they start to age? They have an accident every 50,000 miles when they are brand new and probably maintained by engineers and top level mechanics. What will happen when driven by Joe Six Pack who waits till his brakes make grinding noises and takes the car to Jiffy Lube?

To me it'll all depend on where they are going to and from. You won't get me in any driverless vehicle (electrical or otherwise) that isn't on rails.

Reminds me of the old joke.

The plane had just lifted off the runway and an announcement was made over the speakers to the passengers.

"Ladies and gentlemen this is the first fully automated airplane and has no pilot. We assure you that nothing can possibly go wrong, go wrong, go wrong, go wrong ................."
 
IMO they are a very immature technology being pushed on the public for political reasons.

What too few of the fan boys realize is what they actually "see". A Tesla on autopilot drove straight into the side of a tractor trailer that would have been easily recognized by a human. Now a self-driver Uber vehicle ran over a woman more than 2/3 of the way across a street.

I don't understand how with 100's of race tracks across the USA of various shapes and sizes that sit unused most of the time they are testing these cars on public tax payer funded roads in their current state. The available data on these cars appears limited and vague, definitely not enough for an average person without a technical background to understand all the numerous risks. It does seem these cars were put on public roads without a sliver of what any new concept or development on a vehicle was put through in the past.
 
To me it'll all depend on where they are going to and from. You won't get me in any driverless vehicle (electrical or otherwise) that isn't on rails.

Even rails aren't safe. The Washington DC area Metro train has had a few accidents in recent history where automated safety features have failed. Most all of them had to do with speed control that is supposed to slow the train approaching the stations and curves if the engineer is exceeding safe speed.
 
I don't understand how with 100's of race tracks across the USA of various shapes and sizes that sit unused most of the time they are testing these cars on public tax payer funded roads in their current state. The available data on these cars appears limited and vague, definitely not enough for an average person without a technical background to understand all the numerous risks. It does seem these cars were put on public roads without a sliver of what any new concept or development on a vehicle was put through in the past.

I'm not even close to having my finger on the pulse of what the general public here in America is clamoring for, but having self-driving vehicles on our roads isn't something I want, and I have a difficult time figuring out how any regular traveler might want it. Luckily where I live I don't think it's something I'll see anytime soon, but I digress...

Just seems to me maybe another case of mans reach exceeding his grasp.
 
To me it'll all depend on where they are going to and from. You won't get me in any driverless vehicle (electrical or otherwise) that isn't on rails.

Reminds me of the old joke.

The plane had just lifted off the runway and an announcement was made over the speakers to the passengers.

"Ladies and gentlemen this is the first fully automated airplane and has no pilot. We assure you that nothing can possibly go wrong, go wrong, go wrong, go wrong ................."

every technology goes through its growing pains. This accident will not be repeated. Other than that automation will never stop

dee
;-D
 
There's huge questions behind the "drive" (sorry) to get autonomous cars on the roads. As a person with a minor background in automotive parts design, I get a few SAE magazines, including their newest, Autonomous Vehicle Engineering. and it's got an interesting mix of the "Rah Rah" for new tech, and cautionary tales that many haven't even considered.

Take this article: "Bioprivacy: Designing for a moving target" (which unfortunately I can't find online) which has an expert on such matters considering what happens to all the data that's collected on the human occupant of these cars. Many of them will be equipped to do biometric scanning (retina, fingerprints, etc) for proof of identity before using the vehicle, so now there's a question of who will have access to the data.

Will government agencies have to get warrants before following someone, or will everyone be tracked as a matter of course? This is a huge issue, including the obvious one of having the vehicle redirected to an agency for questioning of the occupants when someone feels the need. Have an outstanding traffic ticket, alimony payment, or similar? Keep your old car...

Along with that, there's the issue of lost jobs and further concentration of wealth. Uber, which is notorious for squeezing their drivers so that their real income is minuscule, wants to go further and eliminate the human driver entirely. That's why they've pushed into AVs, and it was one of their test cars that hit the pedestrian.

For what it's worth, I have questions about how good Uber's technology is, and I'm not the only one. In the same AVE issue I've mentioned, there's an industry research group rating the execution (how good the tech is) and strategy (how they plan to roll out the cars and their robustness of safety and proficiency) of the major players in AV tech.

At the top are Waymo (Google) and GM, just below are Ford, Daimler-Bosch, etc. Where's Uber? At the bottom, just above (surprisingly) Tesla. So one of the most aggressive companies for getting the cars on the road has the worst tech and strategy to do so. And that's with whatever tech Uber was able to hold on to from the theft from Waymo:

New Evidence Could Blow Open the Uber/Waymo Self-Driving Lawsuit | WIRED

So, yes, I expect ultimately there will be a huge presence of self driving vehicles on the road, but right now money is dominating good engineering practice and common sense. This must be addressed or more will die on the roads. Not to say humans are any better, it's scary enough out there as it is.

I also expect that to have a safer environment for self driving cars it will take legislation to get human-driven cars off the roads. This is going to be a massive upheaval not only for car owners, but the car manufacturers. One thing to consider is self driving cars can be "polygamous" and used by dozens of people a day, not just owned by an individual. Again, huge change in dynamics, as far fewer cars will be needed.

Lots more to talk about (this is a subject I'm very interested in), but I need to get some rest after working all night. Perhaps I can get Waymo interested in self-operational design and machining?

[I'm not kidding - same tech will start showing up in manufacturing soon]
 
I think the good idea voters should post their reasoning. I would be curious of your background and experience. Considering that autonomous vehicles share numerous components with your average CNC machine and how just a little moisture or corrosion on a connection can cause one to go haywire I would think your average person with a smidge of tech knowledge would not want to be in the same county as a self driving car.
 
Agree 100% that its a bad idea for all the reasons you mentioned.

Think about what it would take, both from cost and complexity, to add a system to a cnc mill such that you could open the door while its running and toss a chunk of bar stock in the middle of things, and have the machine automatically avoid that obstruction every time, regardless of where the piece lands.

Now, imagine that same system as applied to cars on a mass production basis where the manufacturer wants that same vision and decision making system added to a car that's operating at 5000 ft/min rather than 50 ft/min like the mill. And, oh yeah, make sure it doesn't add more than 50 bucks to the cost of the car.

Couple years ago, my brother in law was driving up a fairly steep 5 mile long grade on I-40 in his Porsche. Road there is divided by the sorta standard concrete divider built from precast sections.

Truck going downhill on the other side loses control and hits the divider. Suddenly he's looking at a 2 ton chunk of reinforced concrete sitting crosswise dead ahead in his lane far too close for him to stop. He jerks the wheel to the right but the car clips the corner of the barrier and flips and comes to a stop on its roof. He's slung around and bruised up a bit, but the EMTs agree with him that he doesn't need to go to the hospital.

So, what does a Uber-bile do in the same situation? Maximum steer input rates have to be controlled such that the vehicle doesn't flip at speed due to trying to dodge a bag of trash in the road. Does it scan the object for size? And then do what? Treat it like the pedestrian, assume it ain't supposed to be there, and slam into it at full speed?

Thanks, but I'll pass for now.
 
From the start I've disliked the idea of self driving cars. Entirely to many variables for a rudimentary system to cope with. In a controlled environment with a road system designed to work with said car... Maybe. But to unleash a driverless car on our roadways, ridiculous. There are less then 1% that are on the roads, much less than, had there been just 20% there would have been mayhem already. I've got a sneaking suspicion that first time drivers and drunks are safer than the current crop of self driving cars. Had that self driver that ran over the pedestrian been someone who had a drink you can get your ass they would have been charged. (Not condoning diving under impairment but that's about the level of ability I consider these self driving cars) rant over.
 
Probably the same sort of arguments back in the days of horses and early cars. Those cars will never replace a horse and and buggy . . .

It's a matter of time.

Seems that autonomous vehicles are already likely safer than drivers. Drivers sometimes die mid-turn. They more often are pissed off, sleepy, drunk, drugged, distracted, texting, eating and applying makeup at 70mph, etc. If humans were all that good, the actual human "emergency backup driver" in the Uber car would have avoided the pedestrian.

As for why they don't test them on closed tracks -- they do. They've been there, done that. Problem is a test track doesn't present the real world complexities that engineers need to encounter, to fix.

So, this really isn't a binary "good idea" vs. "bad idea" question. It's a question of where and how quickly. I suspect lots of us old farts will be driving their own cars for decades ahead -- and mowing down our own share of pedestrians along the way. In my family, my wife can hardly wait until her car drives her around. She's seen her older friends lose their ability to drive -- and hopes to have freedom of movement well into the 90-100 years old most in her family seem to live. I'm kind of hoping it's a long while before we pony up the $$$ for that -- but also that I live long enough to see it.

Meanwhile, we still measure our cars with horsepower -- and worry about today's metaphorical equivalent of shit coming out the tailpipe.
 
I have a friend in Phoenix Arizona that tests driverless cars for a living. IMO, I’d never ride in one or buy one, ever.
Damn govt wants to take away peoples freedom with driverless cars.
 
I recently had an experience while driving that leads to even more questions about the capabilities of SDVs vs human drivers.

While driving at about 35 mph I saw a guy come striding briskly down the sidewalk and then adjust his posture as if he were going to cross the street. I already was gently braking when he resumed walking along the sidewalk but how in hell is a machine to appreciate the subtlety of situations like this? We humans process enormous amounts of data while making decisions and we can notice small clues that a machine might miss.

You might be able to teach the machine the basics of "behind a rolling ball comes a running child" but how do you teach it to realize that a pedestrian might chase a windblown piece of paper into the street? A while back while out for a walk I saw a guy run into the street without thinking after his dog slipped the leash and ran away. The dog had already safely crossed before the guy stepped off the curb right in the path of a car. The driver must have already been braking when he saw the dog and was being cautious in case the dog ran back suddenly. A machine would note that the dog was no longer a threat and proceed normally. And if you taught it to do otherwise travel would be incredibly slow as the SDV imitated those "nervous Nellies" who brake every few seconds because they view everything as a potential threat.
 
Imho there are a few people at UBER who see the accident, rather than as a tragedy, see it as a 'data point'

They'll tweak the code, send the car out, and it'll happen again and again, by the time they've killed maybe 10+ pedestrians, they'll have a vehicle that will be safe to drive on it's own.
 
I guess I look at this alittle differently. Would I prefer to be surrounded by self driving cars or humans on their cellphones? Maybe your area is different ,but as I drive down the highway, I look around and almost 75 to 85% of the drivers I see are on their damn phones. This is my reality. Has anybody watched the dashcam of the lady killed? In the dark, no reflective gear, not in a cross walk, across 40 mph traffic. I agree some super radar sensor tech might have prevented it, but even Jackie Stewart would've blasted her.
 
I guess I look at this alittle differently. Would I prefer to be surrounded by self driving cars or humans on their cellphones? Maybe your area is different ,but as I drive down the highway, I look around and almost 75 to 85% of the drivers I see are on their damn phones. This is my reality. Has anybody watched the dashcam of the lady killed? In the dark, no reflective gear, not in a cross walk, across 40 mph traffic. I agree some super radar sensor tech might have prevented it, but even Jackie Stewart would've blasted her.
I saw the video and it looked like the headlights were either very dim or aimed at the ground. The person was not seen until very close to the car. Also if the car is using the cameras to see things out front then if the camera is using headlight light to sent info for processing..... How many of you have tried to take a photo in a dusk type situation where the picture comes out very dark. LOTS of room for improvement here.
 








 
Back
Top