Jump to content
Twins Daily
  • Create Account

Autonomous Vehicles


Vanimal46

Recommended Posts

Posted

 

I try to ask everyone I know with small children this question... "Do you think your kid will ever have to learn how to drive a vehicle?" The answers are quite interesting.

Kids today don't value driving on their own as much as we did at their age. They would much rather be driven around by someone or something else so they can continue scrolling through their Instagram feed.

 

Yes,

 

I've followed the technology plenty as someone whose family farms and utilizes GPS technology significantly in that industry (as most do now). The average farmer has more screen time per day during planting season than a teenager...scary, but true.

 

The technology has come a long ways over the last decade, but there absolutely still needs to be someone in the vehicle, and not someone who is passively in the vehicle either. Quite frankly, the GPS, depending on weather, software, and all sorts of variables, struggles to make a straight line through an open field at times. The intention of focusing satellite coverage for such technology by population density won't change anything around here for sure, as those same satellite outages and "blinks" that are seen by farmers now would be present in cars. No way could you drive from Sioux Falls to Fargo and expect no issues, let alone getting off interstate to go down gravel roads.

 

Then you have the case of right now, when a road may be okay 10 minutes ago, but melting and culverts opening up around the area are leading to road washouts that are momentary. They've seen testing on farm equipment completely miss such patterns. I'd rather not have my vehicle drive me straight into a stream that is rushing over the road.

 

So, yes, yes, my 14 year-old will definitely need to know how to drive, and absolutely my 4-7 year olds will also need to know.

  • Replies 128
  • Created
  • Last Reply
Posted

 

Uber is scaling back because, like all things Uber does, their testing was reckless and half-assed. It got someone killed.

Google, GM, and VW are not Uber.

 

If I were Google and GM, I'd be offended at being grouped with VW, who is a lot closer to Uber than most want to give credit.

Posted

Yes,

 

I've followed the technology plenty as someone whose family farms and utilizes GPS technology significantly in that industry (as most do now). The average farmer has more screen time per day during planting season than a teenager...scary, but true.

 

The technology has come a long ways over the last decade, but there absolutely still needs to be someone in the vehicle, and not someone who is passively in the vehicle either. Quite frankly, the GPS, depending on weather, software, and all sorts of variables, struggles to make a straight line through an open field at times. The intention of focusing satellite coverage for such technology by population density won't change anything around here for sure, as those same satellite outages and "blinks" that are seen by farmers now would be present in cars. No way could you drive from Sioux Falls to Fargo and expect no issues, let alone getting off interstate to go down gravel roads.

 

Then you have the case of right now, when a road may be okay 10 minutes ago, but melting and culverts opening up around the area are leading to road washouts that are momentary. They've seen testing on farm equipment completely miss such patterns. I'd rather not have my vehicle drive me straight into a stream that is rushing over the road.

 

So, yes, yes, my 14 year-old will definitely need to know how to drive, and absolutely my 4-7 year olds will also need to know.

I don't think autonomous vehicles rely on GPS.

I think it's more scanning and mapping of where things actually are in real time than it is being programmed where things are supposed to be.

Posted

 

I don't think autonomous vehicles rely on GPS.
I think it's more scanning and mapping of where things actually are in real time than it is being programmed where things are supposed to be.

 

And scanning and mapping is done by what technology?

 

It's the same satellite technology, with a different purpose.

 

If you're considering utilizing drone tech, then you've got state drone laws to get through. And all of that is still going to have enough delay that my vehicle will not "know" there's a road washout ahead that just happened a few minutes ago...

Posted

I don't think so. Just eliminating the accordion effect

Just to geek out a little further, a productive way to study traffic is to look at it as a fluid. (I notice that the link you later provided mentions "waves".) Think about how a stream behaves when an obstacle like a large rock is in its midst, with areas of stagnant water resembling certain experiences we've all had in traffic when someone has broken down in a travel lane. Cars can't swirl much, though, except in really dire circumstances.

 

Fluid Dynamics requires calculus but is well worth the years of study to master it. :)

 

 

/ Spoken by someone who failed to master it - I never did quite understand why the airplane wing doesn't just fall down. But I'm told it's way worth it.

Posted

 

See my link.
The University of Illinois study found that even just 5% autonomous cars is enough to virtually eliminate the accordion effect.

 

That article uses the word "could" in a few key spots, which always bothers me.  It's one of those double-speak words that allows statements to be made without really being made.  They also admitted that it was done on a simple round track. They also raved about how surprised they were about how well it worked without really explaining why.

 

The supporting video really only shows the 100% solution in detail.  

 

The robots won't gawk, so they do have my support there.

Posted

 

I don't think autonomous vehicles rely on GPS.
I think it's more scanning and mapping of where things actually are in real time than it is being programmed where things are supposed to be.

 

But they do need GPS to know where to go

Posted

This is a great link. Thanks for posting it.

I mean, those cars aren’t going anywhere, just driving around in circles. Literally.

 

It’s a start, but I’m no less skeptical. Braking does create an accordion effect, but so do intersections, hazards, and reaching destinations. I’d like to see a study that takes those into account.

Posted

And scanning and mapping is done by what technology?

 

It's the same satellite technology, with a different purpose.

 

If you're considering utilizing drone tech, then you've got state drone laws to get through. And all of that is still going to have enough delay that my vehicle will not "know" there's a road washout ahead that just happened a few minutes ago...

Yes, it will know that. It already can do that.

 

The mapping data that it does use, is preloaded onto the vehicles hard drive. It doesn't use satellite or GPS in real time, because obviously, as you said, that would be prone to outages.

 

However, it doesn't rely on mapping data exclusively. Because maps have errors, and autonomous vehicles need to be precise.

 

The technology it would use to know the road is washed out would be its LiDAR system. It would know the road is washed out before a human driver would.

It could possibly also detect it via its radar system, and/or camera system, but the LiDAR system would likely detect it first.

 

Edit: I should clarify, it does use GPS in real time, but doesn't rely on it for precision.

Posted

 

Yes, it will know that. It already can do that.

The mapping data that it does use, is preloaded onto the vehicles hard drive. It doesn't use satellite or GPS in real time, because obviously, as you said, that would be prone to outages.

However, it doesn't rely on mapping data exclusively. Because maps have errors, and autonomous vehicles need to be precise.

The technology it would use to know the road is washed out would be its LiDAR system. It would know the road is washed out before a human driver would.
It could possibly also detect it via its radar system, and/or camera system, but the LiDAR system would likely detect it first.

 

All while one of the leading innovators in the arena completely trashes that technology: https://techcrunch.com/2019/04/22/anyone-relying-on-lidar-is-doomed-elon-musk-says/

Posted

Yes,

 

I've followed the technology plenty as someone whose family farms and utilizes GPS technology significantly in that industry (as most do now). The average farmer has more screen time per day during planting season than a teenager...scary, but true.

 

The technology has come a long ways over the last decade, but there absolutely still needs to be someone in the vehicle, and not someone who is passively in the vehicle either. Quite frankly, the GPS, depending on weather, software, and all sorts of variables, struggles to make a straight line through an open field at times. The intention of focusing satellite coverage for such technology by population density won't change anything around here for sure, as those same satellite outages and "blinks" that are seen by farmers now would be present in cars. No way could you drive from Sioux Falls to Fargo and expect no issues, let alone getting off interstate to go down gravel roads.

 

Then you have the case of right now, when a road may be okay 10 minutes ago, but melting and culverts opening up around the area are leading to road washouts that are momentary. They've seen testing on farm equipment completely miss such patterns. I'd rather not have my vehicle drive me straight into a stream that is rushing over the road.

 

So, yes, yes, my 14 year-old will definitely need to know how to drive, and absolutely my 4-7 year olds will also need to know.

Thanks for that perspective.
Posted

But they do need GPS to know where to go

Yes, and no.

It uses GPS to know where it is in a general sense. It has much more precise systems that adjust and compensate for it's actual exact location.

Do you guys actually think all these smart people poured billions of dollars into this without asking "what happens if the GPS signal goes out?"

Posted

All while one of the leading innovators in the arena completely trashes that technology: https://techcrunch.com/2019/04/22/anyone-relying-on-lidar-is-doomed-elon-musk-says/

He says anyone who "relies" on LiDAR is a fool.

That's a strawman, nobody does and nobody will. As I stated above, other systems like camera and radar (or in Tesla's case) ultrasonic sensors will also detect the washed out road.

It's all about building layers of redundancy. Tesla replaces LiDAR with ultrasonic sensors.

Of course an entrepreneur trashes the main difference between his technology and his rivals. That tactic dates back to Edison/Tesla, probably further.

Posted

...and when those washed out roads now have load upon load of soft gravel, making each 100 feet different than the last?

 

I've lived all my life driving gravel roads through some nasty stuff, and I felt unsafe driving one of the main roads locally over the weekend due to the issues in road quality. There are so many things that play into this beyond just traffic.

 

Sharing the road with a 12-wheel tractor pulling an air seeder

 

Driving a wet gravel road just after a tractor has left significant ruts in various points that could do significant damage to the underside of the car

 

God forbid the various differences in open range laws from state to state that could change the software's response to animals on the road.

Posted

And scanning and mapping is done by what technology?

 

It's the same satellite technology, with a different purpose.

 

If you're considering utilizing drone tech, then you've got state drone laws to get through. And all of that is still going to have enough delay that my vehicle will not "know" there's a road washout ahead that just happened a few minutes ago...

Three things are used:

 

Satellite maps

GPS

Scanning hardware in car (far and away the most important thing)

 

Which is why Tesla won’t get to fully autonomous in their current fleet. They use cameras. Cameras are bad.

 

The rest of the industry is using LIDAR, which is much better than a camera, as it cannot be tricked. An object is there or it is not in the eyes of LIDAR.

 

GPS doesn’t need to be precise if your car has its own set of eyes and makes its own decisions, which is what is being worked on in autonomous cars.

Posted

...and when those washed out roads now have load upon load of soft gravel, making each 100 feet different than the last?

 

I've lived all my life driving gravel roads through some nasty stuff, and I felt unsafe driving one of the main roads locally over the weekend due to the issues in road quality. There are so many things that play into this beyond just traffic.

 

Sharing the road with a 12-wheel tractor pulling an air seeder

 

Driving a wet gravel road just after a tractor has left significant ruts in various points that could do significant damage to the underside of the car

 

God forbid the various differences in open range laws from state to state that could change the software's response to animals on the road.

I guess I don't understand why you think the technology being used would have difficulty reacting and adjusting to those conditions?

It'd actually be better suited than a human, because it would see them from further away, and would analyze exactly how much clearance is needed to clear those ruts, rather than just making an educated guess like a human would.

Posted

He says anyone who "relies" on LiDAR is a fool.

That's a strawman, nobody does and nobody will. As I stated above, other systems like camera and radar (or in Tesla's case) ultrasonic sensors will also detect the washed out road.

It's all about building layers of redundancy. Tesla replaces LiDAR with ultrasonic sensors.

Of course an entrepreneur trashes the main difference between his technology and his rivals. That tactic dates back to Edison/Tesla, probably further.

And Musk is flat-out wrong about LIDAR.

 

Which is why pretty much everybody else is fully on board with the tech. Tesla is the one doing it backwards, not everyone else.

 

I like Musk and all he has brought to technology but the dude is a blowhard.

Posted

 

Three things are used:

Satellite maps
GPS
Scanning hardware in car (far and away the most important thing)

Which is why Tesla won’t get to fully autonomous in their current fleet. They use cameras. Cameras are bad.

The rest of the industry is using LIDAR, which is much better than a camera, as it cannot be tricked. An object is there or it is not in the eyes of LIDAR.

GPS doesn’t need to be precise if your car has its own set of eyes and makes its own decisions, which is what is being worked on in autonomous cars.

 

...and "worked on" is the key, key phrase.

 

GPS technology was dominant in new tractors rolling out before I graduated high school. That was over 20 years ago, and they're still working on the technology due to plenty of different environments in agriculture making it not perfect for all users and requiring more tweeking.

 

That's my exact point. There are a ton of variables present in non-metro settings that will take work to develop, test, and reconfigure. That would make an autonomous car in rural South Dakota a risky proposition until the technology has been truly worked out, and especially for me as a parent to put my children into.

 

Even being kind and saying the autonomous vehicle push will move faster than the worldwide agricultural market push on GPS has, giving 15 years to get technology adequate to be safe on rural roads means my youngest is 19.

 

That still means that yes, they all have to learn how to drive.

Posted

Agreed there. We are likely 20+ years away from 100% fully autonomous. So yeah, anyone already born will likely need to learn how to drive.

 

The better question would have been, will your grandchildren have to? I think there is a good chance that anyone born in 2030 or later will be about as likely to ever drive a car as a teenager now is to ever use a payphone.

Posted

I do understand skepticism toward inevitable advances though, and I do think this one is inevitable.

 

Before the train was invented, nobody had ever traveled faster than a horse could take them.

When trains were in their infancy, many people, even doctors, thought that the g forces created by moving that quickly would kill a human being.

 

Within 100 or so years, we went from nobody ever, in the history of mankind, traveling faster than 30 mph, to people traveling to space at 18,000 mph.

 

When the ball gets rolling on something like this, it often moves much quicker than most people imagined.

Posted

 

I do understand skepticism toward inevitable advances though, and I do think this one is inevitable.

Before the train was invented, nobody had ever traveled faster than a horse could take them.
When trains were in their infancy, many people, even doctors, thought that the g forces created by moving that quickly would kill a human being.

Within 100 or so years, we went from nobody ever, in the history of mankind, traveling faster than 30 mph, to people traveling to space at 18,000 mph.

When the ball gets rolling on something like this, it often moves much quicker than most people imagined.

 

fly? In the sky? ridiculous!

 

Not to mention all those infectious germs in outer space that will kill us.

Posted

Yeah, that's my main point. The technology RIGHT NOW has not shown to be effective in a rural environment. That's going to change, and I'm eager to see it do so, but in the Van example of my children, they will most certainly need the skills to operate a motor vehicle.

 

It is akin to electric cars. It is still quite cost prohibitive (and even less effective) to utilize a vehicle powered with no fossil fuels at all in many areas and in many vocations. That time is coming (sooner than fully automated), and it will be welcomed as well.

Posted

 

Yeah, that's my main point. The technology RIGHT NOW has not shown to be effective in a rural environment. That's going to change, and I'm eager to see it do so, but in the Van example of my children, they will most certainly need the skills to operate a motor vehicle.

 

It is akin to electric cars. It is still quite cost prohibitive (and even less effective) to utilize a vehicle powered with no fossil fuels at all in many areas and in many vocations. That time is coming (sooner than fully automated), and it will be welcomed as well.

 

The technology is here now... and it works. There's a lot of testing to do, but there seems to be an either/or here that says you're either letting the car drive or doing it yourself. I highly doubt it will be completely implemented in that way, for no other reason than that whole MCAS debacle with boeing. There will likely be a way to manually override if needs be.

 

But yeah, computers can drive much better than humans. That's pretty much proven at this point. Whether you want to do it, I get it... but like others have pointed out, you don't have to buy a car that can do that...

 

The other real thing not being mentioned here is the concept of remote work... that's something else that there seems to have been a huge resistance to. Companies are starting to figure that part out... which will also do a ton to get rid of daily commutes.

Posted

The technology is here now... and it works. There's a lot of testing to do, but there seems to be an either/or here that says you're either letting the car drive or doing it yourself. I highly doubt it will be completely implemented in that way, for no other reason than that whole MCAS debacle with boeing. There will likely be a way to manually override if needs be.

 

But yeah, computers can drive much better than humans. That's pretty much proven at this point. Whether you want to do it, I get it... but like others have pointed out, you don't have to buy a car that can do that...

 

The other real thing not being mentioned here is the concept of remote work... that's something else that there seems to have been a huge resistance to. Companies are starting to figure that part out... which will also do a ton to get rid of daily commutes.

Except most homes don't have room for two offices. It's like no one thinks about the need for a space to work.... Not saying you haven't, but I keep reading we'll all work from home .... That's just a way to cut costs for dinner companies, and pass them to employees...

Posted

The other real thing not being mentioned here is the concept of remote work... that's something else that there seems to have been a huge resistance to. Companies are starting to figure that part out... which will also do a ton to get rid of daily commutes.

That's a whole other can of worms ... Getting rid of daily commutes is only one issue ... Not all employees are trustworthy enough to work from home ... There are also often collaboration issues with the younger and older employees in that situation ... It shouldn't be complicated to move that way, but unfortunately, it is.

Posted

 

The technology is here now... and it works. There's a lot of testing to do, but there seems to be an either/or here that says you're either letting the car drive or doing it yourself. I highly doubt it will be completely implemented in that way, for no other reason than that whole MCAS debacle with boeing. There will likely be a way to manually override if needs be.

 

But yeah, computers can drive much better than humans. That's pretty much proven at this point. Whether you want to do it, I get it... but like others have pointed out, you don't have to buy a car that can do that...

 

The other real thing not being mentioned here is the concept of remote work... that's something else that there seems to have been a huge resistance to. Companies are starting to figure that part out... which will also do a ton to get rid of daily commutes.

 

Or you could live in a state that actively discourages it...

Posted

A very good article from a site where I get a lot of my info on autonomous driving.

https://arstechnica.com/cars/2019/04/teslas-autonomy-event-impressive-progress-with-an-unrealistic-timeline/

Great read. Also, the story is actually from today, the 25th, so, timeliness.

 

Reading this, I still don’t have to squint hard to be skeptical that fully autonomous vehicles will arrive any time soon. I’ve already placed my bet on “never,” so I’ll happily hold that bet, but maybe hedge it with “a couple decades” bet too...

 

I really like the Waymo-Tesla comparison in here, seems very much like a tortoise-and-hare scenario, with Tesla going for flash and Waymo playing the long game to get it right. I think having actual people themselves driving millions of miles in these vehicles, people driving them with the autopilot disabled or in shadow mode, is the way to make this last mile or long tail work. (I’m surprised and disappointed the writer sees this as a bad thing, however.) hell, being a Waymo driver could be a career in itself for some people, if the company is truly committed and wants to build good will with these things.

Posted

 

Great read. Also, the story is actually from today, the 25th, so, timeliness.

Reading this, I still don’t have to squint hard to be skeptical that fully autonomous vehicles will arrive any time soon. I’ve already placed my bet on “never,” so I’ll happily hold that bet, but maybe hedge it with “a couple decades” bet too...

I really like the Waymo-Tesla comparison in here, seems very much like a tortoise-and-hare scenario, with Tesla going for flash and Waymo playing the long game to get it right. I think having actual people themselves driving millions of miles in these vehicles, people driving them with the autopilot disabled or in shadow mode, is the way to make this last mile or long tail work. (I’m surprised and disappointed the writer sees this as a bad thing, however.) hell, being a Waymo driver could be a career in itself for some people, if the company is truly committed and wants to build good will with these things.

I've long believed Waymo's approach is the most disciplined and likely to work in the industry. They're using a very principled (and safe) approach.

 

But I think it's a bit silly to say "never". If you realized how quickly this tech and the hardware powering it has advanced, you'd realize that within a decade, the neural processors powering these cars are going to be absolute beasts.

 

You can do a hell of a lot if your processing power jumps twenty fold in a few years.

Posted

 

I've long believed Waymo's approach is the most disciplined and likely to work in the industry. They're using a very principled (and safe) approach.

 

But I think it's a bit silly to say "never". If you realized how quickly this tech and the hardware powering it has advanced, you'd realize that within a decade, the neural processors powering these cars are going to be absolute beasts.

 

You can do a hell of a lot if your processing power jumps twenty fold in ten years.

 

Vice News had an intriguing special report recently that led me to a company, and I found a town hall last evening (then got into researching two things and lost the link - sorry!) the company did with truckers, and one of the issues they still are trying to figure out is how to better protect their laser guidance/sensory system while also keeping it in position to be the most accurate that it can be. I do know that was one major issue that took nearly a decade with agricultural GPS to get where it's at now, and even then, it can get knocked out by the right hit to the vehicle by a pheasant or something like that.

Archived

This topic is now archived and is closed to further replies.

The Twins Daily Caretaker Fund
The Twins Daily Caretaker Fund

You all care about this site. The next step is caring for it. We’re asking you to caretake this site so it can remain the premier Twins community on the internet.

×
×
  • Create New...