Autonomous Cars Will Force Changes To Roads

[Reciprocity]

In a vehicle that is capable of Level 5 Autonomy, you wont need road markings to actually be that good. Remember that Level 5 requires the car to be able to drive in snow where no lane markings are visible and a person does not even have to be in the car. How the heck will that work you might ask? Its actually not as complicated as you might think, but its not easy either. The cars will download hi-def 3D map tiles that includes paths which the car can drive in. The car will use GPS to get its general location then it will use that to get the map tiles. Included in the map tiles are objects the system uses as landmarks. Things like signs, traffic lights, lane markings (when the car can see them) and other things that are at least mostly permanent and stationary. Once the system can recognize a landmark, it can then dial in its location in the 3D tile down to 5-10cm of accuracy. At that point, the car just follows the path and doesnt need to see lane markings at all. Now the vision system will still keep an eye on the road, because things change. Think of it as a backup to the path from the hi-def map tile. Radar and Vision is used to identify objects that are not included in the mapping tile. These objects are important to avoid if they are in way and they are physical objects and not just shadows or something that is not in the way. As you drive, the system will download tiles that you might need, in front of you and adjacent roads. This will keep map tiles small enough to download quickly. If you drive a certain path often, the tiles will be stored for longer and updated as they expire. Tiles will also be kept until you need the space as people tend the drive in the same areas. Tiles will be constantly updated as the many thousands of cars see new objects and new landmarks and as landmarks move or disappear.

People often think that Lidar is required and it might help. But Lidar could also cause confusion. If Linda, Vision and Radar all have a different idea of they see, which is right? Sometimes more sensors is not actually better. Actually better sensors are always better. I believe this is why Tesla decided to focus on Vision. It is going to be very hard to get Vision to work, but not so astronomically harder then using 2 lidars and many radars and a bunch of other stuff. Fully redundant systems is also a bit overrated because if someone fails you have to fix it regardless. If the entire vision system fails, the car can sill pull over the side of the road and turn on its hazards, no different then if you get a flat tire. This is not a military application where the car is going to be shot at and it has to function until the mission is complete at all costs. Anyone of the systems could fail and the car should still be able to safely pull over or at the very worst case, stop in place until it can be towed. And of course the instant it fails, Tesla will know and have a ranger in route.

I know this was a long winded response, but Autonomous cars need to be able to drive in all kinds of situations and road markings wont always be good as they are not always good for people. The advantage autonomous cars will have is that they will have driven every road tens of thousands of times when it might be your first time ever on the road. Machine learning is basically a brute force activity and that's why you need so much computing power in the car.

 

[ElectricTundra]

this is just another example why autonomous driving will not be mainstream for a long time. there are so many variables that need to be overcome that it will be many years until full autonomy can take hold.

Insurance may make it happen much sooner. If an AV or whatever variety is xx% less likely to be involved in a crash and so incur all of the cost of the crash then insurance for that vehicle will be much lower. At the same time insurance rates will rise for vehicles deemed more dangerous (e.g., don't have xx AV safety rating).

 

[essmd]

Pardon my unpopular pessimism, but I sincerely doubt that the advancement of Autonomous Vehicles will result in mandates of road design, lane markings, etc... essentially the "tail wagging the dog".

I do not see any municipality catering to the needs of this feature, unless all of a sudden governments in general develop some intelligence that has rarely happened before.

But admittedly, I also feel that for every step closer to approving autonomous vehicles there will be 2 steps backwards, as testing will reveal yet another element that needs to be addressed and resolved, similar to developers yet to discover what they don't even know they don't know.

I think a good example of this is the difficulty in getting AP2 just to get to AP1 parity. I just replaced my AP1 MX with an AP2 model this week, and what a step in to opposite direction. In a few ways it performs better, but in many ways it is way more inaccurate and "skiddish", decelerating in scenarios AP1 never did, and the image representation is ridiculously "squirrely", with the icons of vehicles ahead popping off and on, all over the place, and what happened to the vehicles in adjacent lanes... has this feature been removed on purpose.

I am not saying Tesla or another manufacturer will never get a vehicle approved of full autonomous activity, but it is much further away than what is currently predicted, especially for universal travel. Just review the chart below

Now if all the vehicles could be made autonomous all at the same time, well obviously that is science fiction, as perhaps Level 3-5 may be as well.

So while of course I paid for EAP, no way was I going to pay for FSD when I think it will not be until AP 6 or 8 before it becomes a reality... and no way will Tesla offer a refund, but perhaps maybe a $1000 or so credit to buying the next HW model.

Even if FSD were to become possible with HW2, would any parent let the car take the kids to school itself? Didn't think so.

 

[johnnyS]

There is a freeway under reconstruction/expansion near us. I have noticed the lady in the navigation gets very upset since she thinks we are on dead end residential streets since the roadbed is not where the GPS thinks it should be. I tend to drive toward the left side of lanes on the freeway due to the motorcycle lane splitters. Autopilot weaves around roughly in the center increasing the chance of a Harley dude knocking off the right side rear view mirror. We are a long way from autonomous autos being able to handle all of these situations.

 

[kort677]

Insurance may make it happen much sooner. If an AV or whatever variety is xx% less likely to be involved in a crash and so incur all of the cost of the crash then insurance for that vehicle will be much lower. At the same time insurance rates will rise for vehicles deemed more dangerous (e.g., don't have xx AV safety rating).

you've got things backwards, autonomous driving will need to be perfected before an insurance company would consider the autonomous vehicle to be safer.

 

[ElectricTundra]

you've got things backwards, autonomous driving will need to be perfected before an insurance company would consider the autonomous vehicle to be safer.

Insurance companies are risk managers, that's how they make money. When AV's prove to be safer, involved in fewer crashes, and cause fewer injuries and fatalities then insurance companies will give them appropriately lower rates. They can't afford not to do so because their competition will.

 

[kort677]

Insurance companies are risk managers, that's how they make money. When AV's prove to be safer,

when being the operative word, until that when happens this is just a speculative comment

 

[ElectricTundra]

when being the operative word, until that when happens this is just a speculative comment

Well yes. Isn't the point of this thread speculation? Our attempts at guesstimating what will happen?

 

[rhumbliner]

The cars will download hi-def 3D map tiles that includes paths which the car can drive in. The car will use GPS to get its general location then it will use that to get the map tiles. Included in the map tiles are objects the system uses as landmarks. Things like signs, traffic lights, lane markings (when the car can see them) and other things that are at least mostly permanent and stationary

While in principle I agree with you, I can think of places in North Dakota, South Dakota, and Nebraska where the system would have to distinguish between blades of grass.

 

[Reciprocity]

While in principle I agree with you, I can think of places in North Dakota, South Dakota, and Nebraska where the system would have to distinguish between blades of grass.

Definitely an issue. I think mobileye did an analysis of this and published the findings. It was something like 2 miles at most between landmarks. Mile markers would work as well. The important thing to note is that once the system knows where it is at down to 5-10cm, it can navigate just based on the speed and telemetry from the car. The landmarks are used to confirm the location as you travel, while the 5-10cm could drift over distances greater then a couple of miles without confirmation. Freeways are nice open roads that would be the easiest to keep dialed in because they are mostly straight and the telemetry of the car should be very accurate for at least a couple of miles. GPS is also more consistent out on the freeways. The car can also use cars around it to correct its position in the lanes, though that would be the least reliable but would be good enough to correct GPS that is accurate to 1.5meters down to considerably less then a meter and more then likely it would just be using this calculation as confirmation that it was still on the right pathing from the HD Maps.

 

[McRat]

...
Even if FSD were to become possible with HW2, would any parent let the car take the kids to school itself? Didn't think so.
...

I drove kids to school .75-1 hour each day (double that when I had to pickup) for 15 years straight. I can guarantee a legally blind drunk driver who has never seen a car before drives better than some humans taking their kids to school.

Many of us humans suck at it. We are on our phones constantly no matter how dangerous conditions are, we can't obey traffic signs, we double park, we gridlock intersections, we don't yield to child pedestrians, we are rude to other drivers, we cut lines, we speed by 100% (50mph in a 25mph = 110 mph on a 55 mph 2 lane), and seldom have situational awareness.

I think I'd feel safer with the kids inside a well controlled car that is actually aware of it's surrounding that has locked doors, seat belts, and airbags. More to the point, I KNOW I'd feel safer if my fellow drivers were forced to drive AV in school areas. It's like a freakin' demo derby some days. I am SO glad my youngest is now driving. But only after 2 professional defensive driving closed course training classes besides the required and utterly worthless state requirements. I sat shotgun with each kid for thousands of miles before they were allowed to drive themselves anywhere. Not because I don't trust THEM, it's because we adults are such crappy drivers today, I'm afraid a fellow parent will kill my kids while they are exchanging gossip at 50mph in a 25mph right through a red light or stop sign. Which is sadly all too common.

Many kids walk, bicycle, or take a bus to school (no seatbelts, iffy drivers).

 

[essmd]

you've got things backwards, autonomous driving will need to be perfected before an insurance company would consider the autonomous vehicle to be safer.

That is correct, I can't imagine insurance companies covering a Tesla FSD even if fully blessed by NTSB and all municipalities, without a massive increase in premiums.

I drove kids to school .75-1 hour each day (double that when I had to pickup) for 15 years straight. I can guarantee a legally blind drunk driver who has never seen a car before drives better than some humans taking their kids to school.

Many of us humans suck at it. We are on our phones constantly no matter how dangerous conditions are, we can't obey traffic signs, we double park, we gridlock intersections, we don't yield to child pedestrians, we are rude to other drivers, we cut lines, we speed by 100% (50mph in a 25mph = 110 mph on a 55 mph 2 lane), and seldom have situational awareness.

I think I'd feel safer with the kids inside a well controlled car that is actually aware of it's surrounding that has locked doors, seat belts, and airbags. More to the point, I KNOW I'd feel safer if my fellow drivers were forced to drive AV in school areas. It's like a freakin' demo derby some days. I am SO glad my youngest is now driving. But only after 2 professional defensive driving closed course training classes besides the required and utterly worthless state requirements. I sat shotgun with each kid for thousands of miles before they were allowed to drive themselves anywhere. Not because I don't trust THEM, it's because we adults are such crappy drivers today, I'm afraid a fellow parent will kill my kids while they are exchanging gossip at 50mph in a 25mph right through a red light or stop sign. Which is sadly all too common.

Many kids walk, bicycle, or take a bus to school (no seatbelts, iffy drivers).

Yup... now if all the vehicles on the road were magically FSD and without a human driver... that's a different story.

 

[Stirfelt]

I want to be excited about the future ... I really do. But
The thread title is "Changes to Roads" I'm thinking about the changes to the Department of Highways maintenance crew ...... to keep the lanes painted much more often than currently; not to mention snow removal ...... I'd be concerned about lane closures (for accidents, repairs, mowing the shoulders, debris on the road and etc).
The sensors on the vehicle will need to be in near perfect condition as well, not covered with dirt, road salt and etc.

As a driving aide ... a useful tool ..... I can see the value.

 

[ItsNotAboutTheMoney]

With the biggest one being "hmm... why is there a bike taking up 1/4 of the car lane and what is the correct manner of dealing with it?"

Overtake giving 3 feet of space. That's what the law says here.

 

[NeverFollow]

We need to have regulations to describe where and how long autonomous
will be allowed to stop to pick up or to drop off passengers.

Currently there are a profusion of "black cars" (Uber/Lift) stopped with their blinkers on
and blocking a full traffic lane, even when there is a single lane with a double yellow line
separating the traffic in the middle, or at a street corner stop sign or a traffic light.

Most of the time I noticed that those "black cars" could have stopped using a driveway,
but the driver don't bother doing so, and don't move away even when multiple cars behind
start to use their car horn.

 

[Bangor Bob]

After using AP1 and AP2 for a while, one thing I've noticed is:

* Frequently, towns/states will let lines on roads fade to nothing before painting

Around here they get repainted every spring, but by mid-winter they're gone. The sand and salt which is necessary in the winter is very abrasive, and chews away the lines. FSD is just going to have to deal with it. Even a well-painted retroreflective line can be nearly invisible at night in the rain.

 

[electracity]

After using AP1 and AP2 for a while, one thing I've noticed is:

* Road standards are all over the place
* Frequently, towns/states will let lines on roads fade to nothing before painting
* Some line conventions - like lines disappearing at intersections, exits - will have to be re-thought
* Having more lines, and standardized, will make help make autonomous driving safer.
* Bikes on roads in lanes meant for 1 car, does not make sense, as the driver/AI will need to break law (drift across line) to drive around them.

Any other laws/conventions people see changing because of robot cars driving around?

Autonomous cars will need to handle all of these conditions before level 4/5 is achieved. The roads are not going to substaintially change for the cars.

An autonomous car can't follow line markings as the primary means of placing the car. This is why everyone is working on high definition mapping.

 

[Driver Dave]

The roads are not going to substaintially change for the cars.

I dunno. When all cars have autonomous features, I be surprised if they didn't. Not because it will be a mandate, just because it will start to make sense to standardize some things.

Once you are watching an AI drive, you really start to see all the faults and inconsistencies in our system of roadways, ones that don't need to be there. Which has been interesting for me.

 

[Electroman]

embedded sensors on the road is the way to go..

 

[Algbc]

embedded sensors on the road is the way to go..

That's true, my next door neighbour used to work for minister of road and transportation of Ontario, I have met with him couple of times in regards to installing cat eye (refelctors) on the roads so during the rain you are able to see the road marking better. I also spoke to him in regards to sensors and synconising the traffic light.
He responded although it's a great idea but it will cost a lot of time and money, and the infestucture is not there yet.
If the city install sensors on the road than they will have to get the money form the federal budget with may increase tax.