Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

AI experts: true full self-driving cars could be decades away because AI is not good enough yet

thefrog1394

Member
Dec 15, 2019
56
42
Ohio
Encouragement for pedestrians, especially children, to wear simple beacons on their person while outside near traffic. Pets too. Clearly the AV should have a very low expectation of harm to people without this, but simple extra precautions like this would likely reduce the likelihood by two or three orders. And remember it's not just something to "enable AVs", it's consistent with an overall desire to reduce vehicle/pedestrian accidents - it's not like we are now in good shape in this regard, before putting AVs on the road.
This sounds straight of a dystopian sci-fi novel. Little Jimmy runs out the door to greet Grandma and in his excitement forgets to don his beacon. The Tesla/Uber/Waymo driving Grandma mows him down. Jimmy's family doesn't stand a chance in court against a multinational autonomous vehicle giant and so the death is ruled no-fault. We are humans. We should live in a world designed for humans. Handing over our freedom to move about within our world without electronic beacons is a terrifying future.

Are there places where this could make sense (i.e. limited access highways)? Sure. Construction workers wearing a beacon for their own safety would make a ton of sense. And autonomy only on interstates would itself have a huge impact. Imagine cross-country drives with the ability to sleep. Commutes with the ability to work for 90%. Entire infrastructure could be built around these limitations (i.e. autonomous vehicle friendly destinations with transit access near urban centers). It's effectively the Personal Rapid Transit concept with the flexibility of using some existing infrastructure and personal vehicles.

But no matter how this ends up, we must be sure not to further hand over our neighborhoods and cities to cars. Tesla, Uber, Waymo, etc would happily take us up on allowing exclusive use of America's 3 million square miles of property dedicated to roads if we let them.
 

JHCCAZ

Supporting Member
Supporting Member
Feb 2, 2021
335
560
Tucson
This sounds straight of a dystopian sci-fi novel. Little Jimmy runs out the door to greet Grandma and in his excitement forgets to don his beacon. The Tesla/Uber/Waymo driving Grandma mows him down. Jimmy's family doesn't stand a chance in court against a multinational autonomous vehicle giant and so the death is ruled no-fault. We are humans. We should live in a world designed for humans. Handing over our freedom to move about within our world without electronic beacons is a terrifying future.

Are there places where this could make sense (i.e. limited access highways)? Sure. Construction workers wearing a beacon for their own safety would make a ton of sense. And autonomy only on interstates would itself have a huge impact. Imagine cross-country drives with the ability to sleep. Commutes with the ability to work for 90%. Entire infrastructure could be built around these limitations (i.e. autonomous vehicle friendly destinations with transit access near urban centers). It's effectively the Personal Rapid Transit concept with the flexibility of using some existing infrastructure and personal vehicles.

But no matter how this ends up, we must be sure not to further hand over our neighborhoods and cities to cars. Tesla, Uber, Waymo, etc would happily take us up on allowing exclusive use of America's 3 million square miles of property dedicated to roads if we let them.
An interesting and very strange take. I'd say this is a completely twisted-around and poorly thought-out interpretation of my suggestion, BTW ignoring key stated points that would refute the dystopian scenario.

Consider: If a father added a reflector to his child's bicycle in 1965, was that a legal invitation to run over other kids who didn't have a reflector? Did it mean "handing over our freedom" to move about or to play safely in the evening? Creating a terrifying dystopian world where we were oppressed by safety devices? Of course not; it was clever piece of technology put to constructive use, a simple and inexpensive way to increase the awareness of drivers that others were present. That is exactly, though more effectively and in more diverse situations, what an inexpensive wireless safety beacon would accomplish in a V2X infrastructure.

The dark and dystopian extrapolation that terrifies you is that a new-fangled device, deployed to help protect children and pedestrians, would somehow shift all safety responsibility away from vehicles and drivers, and thus render our cities and neighborhoods to be human-incompatible.

I'd say the opposite. The real goal of widespread autonomy is to have the opposite effect, enabling a drastically-reduced rate of accidents, injury and death, increased independence and societal participation by the disabled, very old and very young, and notably more efficient roadways that reduce the real-estate and costs of our present ever-widening streets and highways.

Elon Musk goes further by the way, proposing to bury a good many roadways. That's another step beyond, enabled significantly by AV tech, that would reclaim cities and neighborhoods from the increasingly vehicle-dominated trend of the past century or more.
 
  • Like
Reactions: rxlawdude

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
Simulations don't make up for lack of edge cases. That's just a known concept in data science / statistical modeling.

Say you are collecting data in where only 2 features matter. Then you can plot all your data points on a 2d plot. Say Waymo collects the green points. That is their known reality. They create simulations that basically fill in all the area (pink). This is useful so that their model doesn't overfit the green points and creates a more stable solution (within the bounds of the green points).

Tesla, by virtue of collecting a lot more data, finds more edge points that are actually weirder and further from the center (norm) than anyone expected. Tesla then adds simulations (blue + pink) to fill in their gaps.

Waymo doesn't even know the brown points existed. The area between the brown and green are not considered by Waymo. They have no simulation data on them. Ergo, when presented these conditions in real life, their model will fail.

Because these models aren't smart, they are just enormous interpolation machines.

View attachment 673678

In reality, the numeric space between the green and brown points might be small, but happen on so many different dimensions, that the actually "volume" (as in 100 dimension volume) that is missed could be enormous.

Further, the power of simulations is amplified by having more unique raw data. The blue points provide more power than simply adding more dense pink points in a confined region.

Again, Waymo will never talk about this even though it is a fundamental concept in statistical modeling.

The problem is, as Andrej has stated in the past and yesterday. They don't do much simulation nor focus on it. Yesterday he said they use 20k test scenarios for simulation validation.

This is about as bare bone as it gets.

You are equating a company who puts 1-3% into simulation versus a company that puts more than 80% into simulation as being the same.
 
Last edited:
  • Like
Reactions: diplomat33

ZeApelido

Active Member
Jun 1, 2016
3,262
28,254
The Peninsula, CA
The problem is, as Andrej has stated in the past and yesterday. They don't do much simulation nor focus on it. Yesterday he said they use 20k test scenarios for simulation validation.

This is about as bare bone as it gets.

You are equating a company who puts 1-3% into simulation versus a company that puts more than 80% into simulation as being the same.

The problem is you don't understand data science [btw note it's "data" science, not "simulation" science for a reason]. You should leave such analysis to people actually experienced in the field.

They don't mention explicitly how much simulation they are using during training (actually Andrej says they have some really good stuff, but he doesn't want to go into it in that talk). They mention what they use for basically unit testing.

Who cares? Test data should have a mix of data that represents what we perceive as challenging edge cases and also data what we think is the accurate distribution of cases for what will be seen in real life. Totally fine to use real data for that.

In fact, test data should mostly be real data.

Why? Because in machine learning, when you build / test your models, you can train on any data set that you like. Add as many augmented / simulated scenes as you want. But during test, you should be evaluating on real data. Otherwise your accuracy metrics get all screwed up and think they are more accurate than they actually are.

I had this happen to me last year when I had a bug in my code that was allowing simulation data to be added to my test data set. And I was getting higher accuracy marks than I expected. Ones that kept increasing as I increased model complexity. Because if you train your model heavily on simulated data and test heavily on simulated data, the model will figure it all out almost perfectly.

And it's known you are not supposed to do that.

So tell me, is Waymo actually reporting model accuracies that heavily rely on simulation data for evaluation? If so, that's a joke. That is a major faux pax.
 
Last edited:

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
The problem is you don't understand data science [btw note it's "data" science, not "simulation" science for a reason]. You should leave such analysis to people actually experienced in the field.

They don't mention explicitly how much simulation they are using during training (actually Andrej says they have some really good stuff, but he doesn't want to go into it in that talk). They mention what they use for basically unit testing.

Who cares? Test data should have a mix of data that represents what we perceive as challenging edge cases and also data what we think is the accurate distribution of cases for what will be seen in real life. Totally fine to use real data for that.

In fact, test data should mostly be real data.

Why? Because in machine learning, when you build / test your models, you can train on any data set that you like. Add as many augmented / simulated scenes as you want. But during test, you should be evaluating on real data. Otherwise your accuracy metrics get all screwed up and think they are more accurate than they actually are.

I had this happen to me last year when I had a bug in my code that was allowing simulation data to be added to my test data set. And I was getting higher accuracy marks than I expected. Ones that kept increasing as I increased model complexity. Because if you train your model heavily on simulated data and test heavily on simulated data, the model will figure it all out almost perfectly.

And it's known you are not supposed to do that.

So tell me, is Waymo actually reporting model accuracies that heavily rely on simulation data for evaluation? If so, that's a joke. That is a major faux pax.

you seem not to have any understanding what so ever on anything that isn’t in the Tesla bubble. This is usually the case with 99% of all Tesla fans. For example, if I asked you did you watch all/any of the dozens of great presentations on the AD CVPR 2021 other than skipping directly to the Tesla timeline? You would say no and so will the thousands of Tesla fans on Reddit,Youtube and Twitter jumping up about it.

This is the problem

First of all: the information detailed how behind Tesla is on simulation “Under Mr. Bowers, the simulation and maps teams are both still in their infancy, said a person familiar with the situation.” (2019).

In 2020 Andrej said that simulation weren’t their focus since they had real world data so they are focusing ALL their resources on their fleet.

The 20,000 sensor simulation for test and validation is proof that they are at where Waymo was in 2014 in simulation. Before each update Waymo runs millions of scenarios and miles of sophisticated SOTA techniques on all stacks at every level.

Listening to Andrej you would think HD map was unscalable. Yet Mobileye has already solved it at world scale and they have been the only one who has.

Tesla worked on it for 1-2 years and gave up. This is why you shouldn’t take one person/one company for all your source of information and ignore info from everyone else. This is how ignorance and misinformation is breeded. Something you have been demonstrating.
 
Last edited:

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
That's what simulations are doing in the 4-D driving space analogous to your 2D image examples. Simulations change where the cars are, different layouts, change colors, change everything, change and create tons of instances of all the variables we know about.

It's the same concept and provides the same benefits.
That is how data augmentation originally started out when training vision-networks as it was the lowest hanging fruit to try to introduce new variations to the data that you want your model to be invariant to. Adding additional scenarios to further this, either via simulations or other approaches like using GANs to create more data to train on, can and is considered as data augmentation.

No simulation and data augmentation are not the same. Its completely different. Anyone who says otherwise would be laughed out of any academic circle or AV company.

Simulation consists of an environment with certain rules where agents can act and react within.
Data augmentation is literally what it says it is. Augmenting data.

  • A GAN is not a simulator.
  • A simulator CAN contain a GAN (example being smart agents using a variety of trained NN to navigate its environment or used for perception sensor modeling)
  • You can learn/build a simulator using a NN (for example MuZero basically creates its own simulator (model) of the world, with its own rules and acts/plans upon it or Pac-Gan, or simulate parts needed for a simulator for example sensors like lidar, camera, weather data and even pedestrian, vehicle, cyclist behavior, etc).
 
Last edited:

thefrog1394

Member
Dec 15, 2019
56
42
Ohio
An interesting and very strange take. I'd say this is a completely twisted-around and poorly thought-out interpretation of my suggestion, BTW ignoring key stated points that would refute the dystopian scenario.

Consider: If a father added a reflector to his child's bicycle in 1965, was that a legal invitation to run over other kids who didn't have a reflector? Did it mean "handing over our freedom" to move about or to play safely in the evening? Creating a terrifying dystopian world where we were oppressed by safety devices? Of course not; it was clever piece of technology put to constructive use, a simple and inexpensive way to increase the awareness of drivers that others were present. That is exactly, though more effectively and in more diverse situations, what an inexpensive wireless safety beacon would accomplish in a V2X infrastructure.

The dark and dystopian extrapolation that terrifies you is that a new-fangled device, deployed to help protect children and pedestrians, would somehow shift all safety responsibility away from vehicles and drivers, and thus render our cities and neighborhoods to be human-incompatible.

I'd say the opposite. The real goal of widespread autonomy is to have the opposite effect, enabling a drastically-reduced rate of accidents, injury and death, increased independence and societal participation by the disabled, very old and very young, and notably more efficient roadways that reduce the real-estate and costs of our present ever-widening streets and highways.

Elon Musk goes further by the way, proposing to bury a good many roadways. That's another step beyond, enabled significantly by AV tech, that would reclaim cities and neighborhoods from the increasingly vehicle-dominated trend of the past century or more.

I would encourage you to look into the history of jaywalking. Streets were once a community space for people (and horses, which are actually quite good at not running over people. the original autopilot?). As cars entered the scene and pedestrian fatalities skyrocketed, the auto industry pushed a new term, "jaywalking", as a way to stigmatize those who didn't cede the streets to cars (jay meant something along the lines of hick back then).

I bring this up because it's an example of what was once an optional safety measure (crosswalks) quickly being pushed by the auto industry as the only way for a pedestrian to exist in a street without being killed. Fast forward 100 years and we have the world we live in today where cars have free reign of nearly every street in the world with pedestrians, cyclists, kids, street vendors, etc fighting for a small sidewalk or gutter on the side of the road or putting their own life on the line to ride alongside 2 ton death machines.

I think autonomous vehicles becoming a reality brings about an inflection point. I completely agree that this tech can and should be used as a means to make our streets safer for all as you say. But I don't think it's a given that it will lead there. There's absolutely a world where cars whizz past at increasing speeds in urban areas using V2V tech to avoid collisions with one-another and becoming even less hospitable to pedestrians and cyclists. And that world may be one that is actively pushed for by some of the largest and most influential companies in the world as they would stand to make tremendous profits. The decisions we make today will likely change how cities are built for the next 100+ years. Let's make sure we use AV tech to create cities that are human-scale and human-focused, not just ones that look like a cool Jetsons future and make a bunch of profit for Tesla, Waymo, and Uber.
 

orion2001

Member
Apr 14, 2021
118
313
NC
Simulation consists of an environment with certain rules where agents can act and react within.
Data augmentation is literally what it says it is. Augmenting data.
Congratulations, you took an extremely narrow and specific definition of the word "simulation" in this context and built a strawman so you could then get on your soapbox. Google the terms "simulation" in conjunction with "data augmentation" and enjoy all the work from various academics who have NOT been laughed out of academia :rolleyes:
 

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
Congratulations, you took an extremely narrow and specific definition of the word "simulation" in this context and built a strawman so you could then get on your soapbox. Google the terms "simulation" in conjunction with "data augmentation" and enjoy all the work from various academics who have NOT been laughed out of academia :rolleyes:

I explained simulation in the context of ML and AV.
None of it leads to Data Augmentation = Simulation.

Googled 'Simulation':

A simulation is the imitation of the operation of a real-world process or system over time.​

Keyword here is OVER TIME and That's exactly what i relayed.

You have environments, you have actors and you have sensors.

Data Augmentation is manipulating a particular data (for example an image of a car) by adjusting it in ways like changing the colors, or adding noise, or filters or realism, or improving it, or creating replicas to it so you can directly use it within your training set.

Simulation is concerned with imitating the entire process and system overtime. So you are not dealing with just images of a car for example. You are imitating an entire sensor and how it changes/functions through time.

Process/System = an entire sensor (lidar, camera, radar data), or an entire behavior of an actor, or an entire environment map with rules.
Overtime = the process/system's (sensor, environment map, actors) evolution as it acts, is acted upon or reacted to.
 

ZeApelido

Active Member
Jun 1, 2016
3,262
28,254
The Peninsula, CA
No simulation and data augmentation are not the same. Its completely different. Anyone who says otherwise would be laughed out of any academic circle or AV company.

Simulation consists of an environment with certain rules where agents can act and react within.
Data augmentation is literally what it says it is. Augmenting data.

  • A GAN is not a simulator.
  • A simulator CAN contain a GAN (example being smart agents using a variety of trained NN to navigate its environment or used for perception sensor modeling)
  • You can learn/build a simulator using a NN (for example MuZero basically creates its own simulator (model) of the world, with its own rules and acts/plans upon it or Pac-Gan, or simulate parts needed for a simulator for example sensors like lidar, camera, weather data and even pedestrian, vehicle, cyclist behavior, etc).

Congratulations, you read some blogs!
 

orion2001

Member
Apr 14, 2021
118
313
NC
That's not simulation, that's data augmentation.
Let me ask you this, which cars in this image (which is a video) is fake? That's how good simulation has gotten.
You can super-impose depth aware smart agents into a video and they will drive and adapt to the other cars and environment in the video and drive realistically and be shaded with the same lighting of the video.

View attachment 672148

<SNIP>

Dude, you literally called the above synthetic insertion of a 3D rendered model as a "simulation". Your words, not mine. And then after I provided a nuanced reply, you selectively quoted my reply to go on your soapbox. So let me include my full quote below, and draw your attention to the part in bold.

That is how data augmentation originally started out when training vision-networks as it was the lowest hanging fruit to try to introduce new variations to the data that you want your model to be invariant to. Adding additional scenarios to further this, either via simulations or other approaches like using GANs to create more data to train on, can and is considered as data augmentation. The purpose of all of these are to ultimately achieve the same thing, which is to add more variety and richness to the source data used for training that you want the network to be invariant to so that it can generalize better across a wider set of inputs in the real-world.

On the other hand, if you consider simulations for training policies and not the underlying perception networks, one could argue that it is not really data augmentation so much as providing truly unique data for unique scenarios to train against. At the end of the day, it is not very meaningful to debate whether something is or isn't data augmentation. The end goal is the same. What really matters is how much "novel" information your data augmentation scheme can introduce to make your networks more robust.

While simulations are a powerful tool and can be very helpful, I don't see them being as useful in capturing the long tail of exceptional/weird events that humans deal with every day without even thinking about it. eg: A large plastic bag flutters across the front of the car and the Lidar freaks out because there is an obstacle in the way when humans know they don't need to worry about it at all and will never hit the brakes in this scenario.
As I mentioned in my quote above, and I will repeat again:

At the end of the day, it is not very meaningful to debate whether something is or isn't data augmentation. The end goal is the same. What really matters is how much "novel" information your data augmentation scheme can introduce to make your networks more robust.

So it would be good to just move on from this pointless discussion and stop selectively quoting parts of someone's replies from several pages back so you can make snide comments and pontificate after reading some blog posts.
 

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
Dude, you literally called the above synthetic insertion of a 3D rendered model as a "simulation". Your words, not mine. And then after I provided a nuanced reply, you selectively quoted my reply to go on your soapbox. So let me include my full quote below, and draw your attention to the part in bold.


As I mentioned in my quote above, and I will repeat again:

At the end of the day, it is not very meaningful to debate whether something is or isn't data augmentation. The end goal is the same. What really matters is how much "novel" information your data augmentation scheme can introduce to make your networks more robust.

So it would be good to just move on from this pointless discussion and stop selectively quoting parts of someone's replies from several pages back so you can make snide comments and pontificate after reading some blog posts.

Realistic Video Simulation via Geometry and Traffic Aware Composition for Self-Driving

Its an actual simulation (imitation of process/system over time) not just an insertion of a picture or 3d model.​

GeoSim also have a separate feature for data augmentation. If video simulation and data augmentation were the same thing they wouldn't have separated them.

How is this impossible for you to understand? Educate yourself!​


Waymo just did a presentation yesterday about simulation and data augmentation and had no problem separating them. Because they are two completely different things. just like the thousands of other actual researchers who do the same. Only ignorant Tesla fans have a problem with it because they have no clue what they are talking about.


So it would be good to just move on from this pointless discussion and stop selectively quoting parts of someone's replies from several pages back so you can make snide comments and pontificate after reading some blog posts.

I didn't just "read some blog posts" But i rather just read some blog posts and be knowledgeable than to be as ignorant as a tesla fan.
 
Last edited:

diplomat33

Well-Known Member
Aug 3, 2017
8,168
9,714
Terre Haute, IN USA
Waymo just did a presentation yesterday about simulation and data augmentation and had no problem separating them.

Speaking of that, Waymo's simulation city intrigues me. They say it will allow much faster FSD training than real world driving. IMO, it could be a game changer.

gZWYftz.png
 

Bladerskb

Senior Software Engineer
Oct 24, 2016
2,429
2,806
Michigan
Congratulations, you read some blogs!
Thanks.. If that's what I did then I rather just read blogs than having no clue and watching Tesla AP day on repeat like most Tesla fans.

Speaking of that, Waymo's simulation city intrigues me. They say it will allow much faster FSD training than real world driving. IMO, it could be a game changer.

gZWYftz.png

Yeah it looks like Waymo is building a WaymoZero City Simulator using their SurfelGan 3D map that they relight with 24 hour realistic Time of Day, weather, seasons with vectorNet, various sensor simulation NNs, and smart Imitation and RL learned agents that Drago have talked about.

A simulator so realistic, not even a human would be able to differentiate it from a real feed from their cameras. A 1:1 scale and match in both geometry and textures would definitely be a game changer.

In their paper last year they said:

"Autonomous driving system evaluation requires the ability to realistically replay a large set of diverse and complex scenarios in simulation capturing sensor properties, seasons, time of day, and weather. Developing simulators that support the levels of realism required for autonomous system evaluation is a challenging task. Furthermore, because the environment we are building is a high-quality reconstruction based on the vehicle sensors, it naturally closes the domain gap between synthetic and real contents, which is present in most traditional simulation environments. "

If they can complete this in the next 2 years, its essentially game over because they are doing this AT scale.
They would completely shut the door of reality gap for good.
They could do Population Based Self-Play Reinforcement Learning using their current AV software as bootstrap with other imitation learned models from their waymo dataset and from what Latent Logic has been doing with traffic cameras.

Doesn't mean they will solve L5 in an instant but they would dramatically improve their system exponentially. A TRUE quantum leap. This is why they said "much faster than real world driving".

This would be an Alpha Zero moment if they can pull this off.

This WaymoZero Sim would be essential for mass copy/pasting as they can create one for different cities. For example deploying in 10 cities in a quarter. Mass copy/paste would begin in 2024 imo.

OwvR1Ns.png
 

orion2001

Member
Apr 14, 2021
118
313
NC
Lol @ the irony of being called a Tesla fan when anyone seeing my post history would find me to be anything but that. I guess sooner or later you find any FSD discussion around here and make it unbearable with personal attacks and incoherent, moving goal post arguments. The regular FSD thread sufferred the same fate. Enjoy your echo chamber after driving everyone else away with the continuous attacks and ramblings.

I'm off to go do the real thing - train a neural net that does object detection, semantic segmentation, object tracking amongst other things... but hey, what do I know about any of this stuff 🤷‍♂️
 

mark95476

Active Member
Jun 21, 2020
1,699
1,036
Bay Area CA
That's their entire intent.

Lol @ the irony of being called a Tesla fan when anyone seeing my post history would find me to be anything but that. I guess sooner or later you find any FSD discussion around here and make it unbearable with personal attacks and incoherent, moving goal post arguments. The regular FSD thread sufferred the same fate. Enjoy your echo chamber after driving everyone else away with the continuous attacks and ramblings.

I'm off to go do the real thing - train a neural net that does object detection, semantic segmentation, object tracking amongst other things... but hey, what do I know about any of this stuff 🤷‍♂️
 
  • Like
Reactions: mikes_fsd

Barklikeadog

Active Member
Jul 13, 2016
1,881
1,366
PA
An interesting and very strange take. I'd say this is a completely twisted-around and poorly thought-out interpretation of my suggestion, BTW ignoring key stated points that would refute the dystopian scenario.

Consider: If a father added a reflector to his child's bicycle in 1965, was that a legal invitation to run over other kids who didn't have a reflector? Did it mean "handing over our freedom" to move about or to play safely in the evening? Creating a terrifying dystopian world where we were oppressed by safety devices? Of course not; it was clever piece of technology put to constructive use, a simple and inexpensive way to increase the awareness of drivers that others were present. That is exactly, though more effectively and in more diverse situations, what an inexpensive wireless safety beacon would accomplish in a V2X infrastructure.

The dark and dystopian extrapolation that terrifies you is that a new-fangled device, deployed to help protect children and pedestrians, would somehow shift all safety responsibility away from vehicles and drivers, and thus render our cities and neighborhoods to be human-incompatible.

I'd say the opposite. The real goal of widespread autonomy is to have the opposite effect, enabling a drastically-reduced rate of accidents, injury and death, increased independence and societal participation by the disabled, very old and very young, and notably more efficient roadways that reduce the real-estate and costs of our present ever-widening streets and highways.

Elon Musk goes further by the way, proposing to bury a good many roadways. That's another step beyond, enabled significantly by AV tech, that would reclaim cities and neighborhoods from the increasingly vehicle-dominated trend of the past century or more.
The amish did make a good case that a modern day reflector was an intrusion on their freedoms. But they lost since driving, or other activities on a public roadway are a privilege and not a right.
 

Products we're discussing on TMC...

About Us

Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.

Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


SUPPORT TMC
Top