Tricking a Tesla to drive with no one in the driver's seat

Amtrak Unlimited Discussion Forum

Help Support Amtrak Unlimited Discussion Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Joined
Apr 5, 2011
Messages
6,090
Location
Baltimore. MD
Tesla Will Drive With No One in the Driver's Seat - Consumer Reports


I know people are eager for self-driving cars, but this seems a little premature.
The features really shouldn't be called "autopilot." This stuff is actually "driver assist," not self-driving.
Tesla Autopilot Crash Highlights Limitations of Driver-Assist Systems (consumerreports.org)

I have a lane departure alert, collision avoidance, and adaptive cruise control on my 2017 Toyota RAV-4. The lane departure alter usually works, though it gets confused by poorly painted lines, and sometimes goes off for no reason. The collision avoidance didn't stop my wife from running into another car that had run a stop sign in front of her. I've seen no reason for using the adaptive cruise control.

True self-driving cars are a long way off, if they ever do become reality on public roads. I wish people marketing driver assistance stuff would be careful about how they describe it. Right now, if you ride in a car on a public road, somebody has to be driving the thing.
 
Last edited:
Tesla Will Drive With No One in the Driver's Seat - Consumer Reports


I know people are eager for self-driving cars, but this seems a little premature.
The features really shouldn't be called "autopilot." This stuff is actually "driver assist," not self-driving.
Tesla Autopilot Crash Highlights Limitations of Driver-Assist Systems (consumerreports.org)

I have a lane departure alert, collision avoidance, and adaptive cruise control on my 2017 Toyota RAV-4. The lane departure alter usually works, though it gets confused by poorly painted lines, and sometimes goes off for no reason. The collision avoidance didn't stop my wife from running into another car that had run a stop sign in front of her. I've seen no reason for using the adaptive cruise control.

True self-driving cars are a long way off, if they ever do become reality on public roads. I wish people marketing driver assistance stuff would be careful about how they describe it. Right now, if you ride in a car on a public road, somebody has to be driving the thing.
There are several Companies testing " Driverless Cars" here in Austin.

Of course there is a "Driver" in the Front Seat on the Passenger side, with Controls similar to those that used to be used to Teach Student Drivers back in the day.

With the traffic the way it is in Major Cities and out on most Interstates, I really don't see the practicality, not to mention the Huge Expense that would be required to implement this technology.

I even think those Commercials that show Cars that Park themselves are an idiotic gimmick!

In Texas, Paralell Parking used to be part of the Drivers Test, if you couldn't do it, you Failed the Test.

Whenever I'm riding in any means of Transportation, I want a Human @ the Controls, not a Computer!( I dont even trust those Airport Trains, (or Monorails) that dont have Drivers!

Auto pilots are OK on Planes, as long as there are 2 Pilots in the Cockpit!😉
 
On the one hand to a typical consumer "autopilot" has a name that implies it already does things it cannot yet do in a reliable fashion at this time. That being said Tesla's Autopilot already does more than the original "autopilot" for aircraft did when those devices were first implemented. Tesla themselves call it a beta product and it is gaining ground at a rapid pace. While it may sound like it serves a similar function Tesla Autopilot is an order of magnitude more advanced and nimble than the basic assist systems found on the likes of Honda and Toyota. Barring government intervention there is almost no way that self-driving vehicles do not become a reality in the next few years. The money to be made and saved is high enough that it will overwhelm almost any other consideration at this point.
 
While Tesla Autopilot is not intended to replace the driver, Tesla has been beta testing Full Self-Driving Mode (FSD) for some time, which is designed to drive the car on its own in urban environments. You can find lots of videos taken from the driver's POV on YouTube; I think Tesla was encouraging beta testers to post them.

I haven't checked them out for the past couple of months, but the last time I did, it was pretty amazing what the on-board computer could do. There was still quite a bit of work needed to handle some situations, like unmarked parking lanes, as an example.
 
On the one hand to a typical consumer "autopilot" has a name that implies it already does things it cannot yet do in a reliable fashion at this time. That being said Tesla themselves call it a beta product and it is gaining ground at a rapid pace. In addition Tesla's Autopilot already does more than the original "autopilot" for aircraft did when those devices were first implemented. While it may sound like it serves a similar function Tesla Autopilot is an order of magnitude more advanced and nimble than the basic assist systems found on the likes of Honda and Toyota. Barring government intervention there is almost no way that self-driving vehicles do not become a reality in the next few years. The money to be made and saved is high enough that it will overwhelm almost any other consideration at this point.
Who is going to put up the Trillions it will take Chris?

The current group of Clowns and Grifters we have in Congress in Washington cant even agree to spend what's necessary to repair/replace the crumbling Infastructure we have now, let alone invest in something as exotic as driverless vehicles!

Having every Billionaire in the World put up their Fortunes still wouldnt produce enough money to pay for this, and also picture the Mess that will occur while the system is being built,tested and all the " bugs" are worked out.

While he may be a Visionary, Elon Musk has also had lots of Failures( ie Space and Fires in his Vehicles,he still hasnt made a Profit yet!) and is somewhat of a Hustler in that he plays Governments against each other as he moves his operations from State to State.( Texas and Austin are his latest " Suckers")

Only the Government would have the means to do this and wouldnt that be called " Socialism" by the Republicans?

YMMV
 
Last edited:
On the one hand to a typical consumer "autopilot" has a name that implies it already does things it cannot yet do in a reliable fashion at this time. That being said Tesla's Autopilot already does more than the original "autopilot" for aircraft did when those devices were first implemented. Tesla themselves call it a beta product and it is gaining ground at a rapid pace. While it may sound like it serves a similar function Tesla Autopilot is an order of magnitude more advanced and nimble than the basic assist systems found on the likes of Honda and Toyota. Barring government intervention there is almost no way that self-driving vehicles do not become a reality in the next few years. The money to be made and saved is high enough that it will overwhelm almost any other consideration at this point.
I'm not a Tesla owner, but I believe there are 2 different products: AutoPilot is the freeway-driving tool which has been in general release for quite a while, and is similar to other automaker's safety suites but perhaps a step more advanced; FSD is the fully automated driving system which is in beta.
 
I agree that the technology is not yet perfect, but it is getting better and better, unfortunately, human driver's aren't doing the same. Already, I think that today's early autonomous cars do a better job than probably more than half the human driver's, and consequently, I would rather drive surrounded by these cars than most human driver's. Eventually, almost all cars will be autonomous, and with advanced telemetry between them, cars will be able to safely operate closer together, and at higher speeds than possible today.
 
Who is going to put up the Trillions it will take Chris? The current group of Clowns and Grifters we have in Congress in Washington cant even agree to spend what's necessary to repair/replace the crumbling Infastructure we have now, let alone invest in something as exotic as driverless vehicles! Not even Jeff Bezos or Warren Buffett have enough money to pay for this, and also picture the Mess that will occur while the system is being built,tested and all the " bugs" are worked out. YMMV
Self-driving does not have to be perfect, it only has to be better than a typical human driver, and that bar is dropping over time as drivers goof around with text messages and even video chat while driving. I honestly cannot believe half the crap I see other drivers doing when I look over to see why they've entered my lane or whatever other nonsense.

I'm not a Tesla owner, but I believe there are 2 different products: AutoPilot is the freeway-driving tool which has been in general release for quite a while, and is similar to other automaker's safety suites but perhaps a step more advanced; FSD is the fully automated driving system which is in beta.
You are absolutely correct, but they've changed the terms and meaning over time, and I would imagine they will eventually go back to calling the whole thing Autopilot 2.0 or whatever at some point. Just a guess though.
 
With the traffic the way it is in Major Cities and out on most Interstates, I really don't see the practicality, not to mention the Huge Expense that would be required to implement this technology.


I think the endgame here is to create the "smart" city. Traffic Jams will probably be a thing of the past because the cars can send info to the traffic systems and they can optimize the flow of traffic better than today. Also, I've seen some people do crazy things behind the wheel. Also it might hinder the people who want to street race. There was a kid here a few years ago killed a woman and her baby doing 100+ in a 40.
 
I'm not a Tesla owner, but I believe there are 2 different products: AutoPilot is the freeway-driving tool which has been in general release for quite a while, and is similar to other automaker's safety suites but perhaps a step more advanced; FSD is the fully automated driving system which is in beta.
Fatal 'Driverless' Tesla Crash | Autopilot - Consumer Reports

According to the Consumer Reports testers, neither of Tesla's Autopilot nor "Full Self Driving" features make a car self driving. Drivers still need to be on board and ready to take over, if necessary. Thus, Tesla is really being irresponsible by using marketing terms that imply that the cares are self-driving.
 
Barring government intervention there is almost no way that self-driving vehicles do not become a reality in the next few years.
Without getting into too many details, let's just say that I am in this space and am privy to a lot of high level analysis. Unfortunately, this is the most pervasive myth that is out there. The odds are that a true, commercially practical, Level 5 autonomous vehicle will NOT be available in the next few years.

The simplest way to explain it is that getting to 90% of where we need to be has been relatively straight forward. It's the last 10% that is proving to be the challenge. It's one thing to do trials with insanely expensive equipment in places like Arizona or California with defined routes. But don't get fooled by those successes. Getting a computer to work like a human brain is a challenge that is incredibly difficult. Are those grass clippings on the road a lawn? What to do with the lines that kids drew on the street in chalk? What is that blowing snow? Humans are incredibly adept at figuring those things out. Computers aren't there yet.

I want to see Level 5 autonomous vehicles as badly as anyone else. But the progress we are making now is incremental. The easy part is past us.

Even Elon Musk, who is known for hype and self-promotion, said this last year: "We believe that market-ready Level 5 vehicles will not be available for several more years and have significant product development hurdles to clear before then."

So is a Level 5 vehicle possible in a few years? Yes. But that is an absolute best case scenario and in all likelihood it will only work in certain situations - and it will not be priced for the masses.

I am just passing on what people who are much smarter than me, and much more knowledgeable about the challenges, are saying.
 
Last edited:
Without getting into too many details, let's just say that I am in this space and am privy to a lot of high level analysis. Unfortunately, this is the most pervasive myth that is out there. A true Level 5 autonomous vehicle is more than "a few years" away - at least if we are talking about something beyond a niche use at extremely high cost. And even if we aren't talking about that, there is a very high likelihood that it will be more than a few years awat,

The simplest way to explain it is that getting to 90% of where we need to be has been relatively straight forward. It's the last 10% that is proving to be the challenge. It's one thing to do trials with insanely expensive equipment in places like Arizona or California. But don't get fooled by those successes. Getting a computer to work like a human brain is a challenge is incredibly difficult. Are those grass clippings on the road a lawn? What to do with the lines that kids drew on the street in chalk? What is that blowing snow? Humans are incredibly adept at figuring those things out. Computers aren't there yet - at least to the extent we need them to be at a price that the masses can afford.

I want to see Level 5 autonomous vehicles as badly as anyone else. But the progress we are making now is incremental. The easy part is past us.

Even Elon Musk, who has every incentive to over-hype the state of progress, said this last year: "We believe that market-ready Level 5 vehicles will not be available for several more years and have significant product development hurdles to clear before then."

So is a Level 5 vehicle possible in a few years? Yes. But that is an absolute best case scenario and in all likelihood it will only work in certain situations - and it will not be priced for the masses.

I am just passing on what people who are much smarter than me, and much more knowledgeable about the challenges, are saying.
Good summary of the 'state-of-the-art'...thanks.
 
It seems to me that the main push towards driverless vehicles is in the field of commercial transportation? Not so much about "better drivers" more about cutting employment costs?
For the last decade or so, it is not so much about cutting costs, but more about finding commercial driver's. There is such a shortage of qualified CDL driver's, that truck and bus carrier's are paying large 'sign-on' bonuses in an attempt to attract recruits.....
 
For the last decade or so, it is not so much about cutting costs, but more about finding commercial driver's. There is such a shortage of qualified CDL driver's, that truck and bus carrier's are paying large 'sign-on' bonuses in an attempt to attract recruits.....
On the other hand, I heard from independent owner-operators how tight the market is, how the rates they get are so low that they can't make a living, etc. Then there are the drayage truck operators (they people who drive the containers from the port to the distribution center) who are apparently really exploited. Aside from the fact that who wants to work tethered to a GPS monitor? I think if they truck lines paid well and had good working conditions, they'd have more people interested.
 
Even Elon Musk, who is known for hype and self-promotion, said this last year: "We believe that market-ready Level 5 vehicles will not be available for several more years and have significant product development hurdles to clear before then."

So is a Level 5 vehicle possible in a few years? Yes. But that is an absolute best case scenario and in all likelihood it will only work in certain situations - and it will not be priced for the masses.

The trouble with Elon Musk is that he might say the right things in some places, but the by marketing his product as a "full self-driving" package he's implying that you can buy a level 5 vehicle today, at least from him.

And you're right, when they are available, I would see them more doing stuff like airport shuttle buses and such, not for the average yahoo to be able to get in a car and spend his time texting.

By the way, as fans of trains and other forms of public transportation, we might not be all that thrilled about the vision of autonomous vehicles that these guys are putting out. If they really do solve the problems of emissions and traffic jams, that will be serious competition with the use of public transit, or even taking a train for longer trips. When I went to industry conferences, I did notice a subcurrent, not usually spoken, about how this technology would make public transit obsolete. I think the auto industry may have some long-term worries about the viability of their product, and this might be one way to shore it up. There are other good reasons for us to reduce our use of private, individual vehicles to maintain our mobility, unfortunately, we are culturally predisposed to wanting private vehicles. (Some of the cultural factors may be related to the distaste many of us have for having a random seatmate when we travel in coach. :) )
 
There have been some very good studies on the impact fully autonomous vehicles will have on certain markets.

Most people think that fewer vehicles will be sold when autonomous vehicles are available. This is because ride-sharing will be much cheaper (with no driver to pay) and families can share cars much easier. For example, I could send my car to drop the kids off at school and then it could come back and pick me up so I can get to work. After it drops me off at work, it can go back home and be used by my spouse during the day.

The research doesn't agree, however. The research predicts some shifts in the market, but that vehicles will still sell very well. The main reasons are: (1) Americans love their cars and want one in their driveway; (2) Waiting for a ride-share vehicle gets old, especially when just want to run to the store for a gallon of milk but you are told that you have to wait for 20 minutes before a ride will arrive. Americans want their instant gratification; (3) People who can't drive will now be able to own and use a car. Demand for vehicles in rural areas will remain very strong. Urban areas may see a softening - but overall the market for vehicles sales will remain robust.

Another huge challenge is integrating fully autonomous vehicles with older rolling stock. One of the best ways to get autonomous vehicles out there in the real world is if they can talk to each other. But there are going to be millions of cars on the road that won't have that capability. It's a chicken and egg problem.

As for impact on public transportation, that is a genuine concern. Keep in mind that once there is widespread adoption of autonomous vehicles, traffic jams will be much less common. It's also a concern for longer trips because cars can be designed to emphasize comfort. The interiors will be very different. Your car will become more like a "pod" in which you can work and sleep. In other words, it will be your own personal roomette that you can take wherever you want - and there will probably be vigorous debates on whether or not it should have a toilet. That said, motion sickness is a genuine problem.

Lastly, there will not be a magical moment when autonomous cars hit the road. Instead, you are going to see technology creep. We are already seeing it with autonomous cruise control, lane keep assist, etc. We aren't going to just jump from Level 1 or 2 to Level 5.
 
Last edited:
On the other hand, I heard from independent owner-operators how tight the market is, how the rates they get are so low that they can't make a living, etc. Then there are the drayage truck operators (they people who drive the containers from the port to the distribution center) who are apparently really exploited. Aside from the fact that who wants to work tethered to a GPS monitor? I think if they truck lines paid well and had good working conditions, they'd have more people interested.
Good points.
There has always been a disparity among commercial driver's. Owner operators, especial 'bedbug haulers' (moving vans), have to virtually live in their vehicles, sometimes for months at a time before ever getting home, in order to make their payments on their rigs. Most of these rigs have what amounts to a camper on the tractor. It used to be, a Teamster's Union driver working for a motor freight common carrier, such as the old Consolidated Freightways, ran their driver's relaying trucks over long distance's, like a Greyhound bus driver. Some of them went from places like Cheyenne to North Platte, and then back again in the same day, sleeping in the comfort of their own bed each night. And they received excellent pay and benefits. But nowadays, carrier's don't want to employ driver's, they would rather they work for them as "independent contractor's".
 
Unfortunately, this is the most pervasive myth that is out there. The odds are that a true, commercially practical, Level 5 autonomous vehicle will NOT be available in the next few years.
I believe that qualified autonomy in the form of commercial vehicles operating through approved routes and conditions without human drivers meets a definition of “self-driving” that can be achieved in the next few years.

Getting a computer to work like a human brain is a challenge that is incredibly difficult.
Which is why part of the breakthrough will involve modifying some roads to function more like a computer expects. Until now nearly all roads were designed and maintained at a level appropriate for fuzzy human logic, but instead of bringing every vehicle all the way up to Level 5 why not bring some routes down to Level 4 instead?

Humans are incredibly adept at figuring those things out. Computers aren't there yet.
Humans are worse at following the rules but better at anticipating what other human drivers will do around them.

NOTE: My post is intended as a generalization and the terminology I’ve used is imprecise, but I do believe there is an achievable middle ground where technical ability and financial benefit will meet over the next few years.
 
I believe that qualified autonomy in the form of commercial vehicles operating through approved routes and conditions without human drivers meets a definition of “self-driving” that can be achieved in the next few years.
What you described is not a Level 5 autonomous vehicle.

Which is why part of the breakthrough will involve modifying some roads to function more like a computer expects.
We haven’t even agreed on a standard even if we wanted to do that. It’s more than a few years off, other than the most limited of demonstrations.

I appreciate your optimism, but the experts are quite clear that the pace of innovation has slowed substantially. In the next few years we will see more features, but we won’t see a true Level 5 vehicle. We also won’t see widespread adoption of autonomous road infrastructure. Trust me. I have read the studies, some of which were privately commissioned.

We will get there, however.
 
I appreciate your optimism, but the experts are quite clear that the pace of innovation has slowed substantially. In the next few years we will see more features, but we won’t see a true Level 5 vehicle. We also won’t see widespread adoption of autonomous road infrastructure. Trust me. I have read the studies, some of which were privately commissioned.
I agree that universal autonomy, where you can pick any address you please and a consumer grade vehicle will reliably find its way safely under any conditions without monitoring, is indeed a long way off. I do believe there is a substantial middle ground that does not require that level of sophistication to achieve self-driving through special industrial areas and freight corridors, assuming regulators and insurers are on board. This is just my opinion though.

Lastly, there will not be a magical moment when autonomous cars hit the road. Instead, you are going to see technology creep. We are already seeing it with autonomous cruise control, lane keep assist, etc. We aren't going to just jump from Level 1 or 2 to Level 5.
👍
 
Last edited:
Here is an interesting video I just stumbled upon, which has quite a bit of bearing on the subject matter of this thread, even though it is about a Waymo funded Veritasium puffery about Waymo with an attempt to posture as if it is a balanced educational presentation, and does not involve Tesla.



As usual sociological manipulation using the internet can be pernicious, and one needs to be on the lookout for such.
 
One interesting wrinkle in the eventual adoption of autonomous vehicles is the reaction that people have to harms caused by them. In an experiment sponsored by an insurance company, subjects were given one of two descriptions of an accident. In both variations of the story given to subjects, the consequences were identical--two people killed. But in one of the variations, the car causing the accident was being driven by someone who changed lanes without looking first--clear negligence by that driver. In the other, a computer program in the car was incorrectly programmed, causing the accident. In the computer-caused-crash, the subjects were told that the automous driving program that led to the crash would, however, have prevented half of all crashes caused by human driver negligence. (That statistic is of course made up--since the programs don't currently exist, we have no way of knowing how much better they might be than the population of human drivers.) But, the takeaway was the the subjects, acting as mock 'jurors,' assessed far higher damages when the accident was caused by a faulty program than they did to the accident by the negligent driver.

Insurance companies are trying to figure out how to price policies for autonomous vehicles once they get created--it can't be done simply by looking at the actuarial likelihood of accidents, it has to also take into account the human jurors' tendency to more harshly treat computer-caused harm than human-fault-caused-harm. Apparently we think, "to err is human," but computers better be close to divine...
 
Here is an interesting video I just stumbled upon, which has quite a bit of bearing on the subject matter of this thread, even though it is about a Waymo funded Veritasium puffery about Waymo with an attempt to posture as if it is a balanced educational presentation, and does not involve Tesla. As usual sociological manipulation using the internet can be pernicious, and one needs to be on the lookout for such.
The Veritasium video could and should be questioned and challenged, but this supposedly balanced and truth-focused author gleefully interprets every aspect in the most absurd manner possible. Vague body language and casual analogies are deemed highly deceptive but countered with flimsy "some people say" accusations that reframe the debate as "pipe dream or highly dangerous?" Driving is already dangerous, in fact it's the most dangerous thing most of us do on a regular basis, and adding autonomous vehicles into a dangerous situation will always be unsafe to one degree or another.

Autonomous vehicles need more flexible solutions and more predictable conditions to achieve extremely reliable safety standards, but the amount of effort required to make driving safe for human drivers likely dwarfs the effort required to make automatic driving substantially safer than human driving. Unfortunately instead of using his profound insight to suggest alternative solutions to autonomous driving problems, the author seems content to revel in the role of spoiler. His half-hearted and tacked-on appeal for more public transit notwithstanding. In other words he chooses to fight sponsored puffery with obnoxious pettiness.

In the end the main takeaway is that Tom Nicholas hates sponsored content, even when clearly identified, but has no problem telling us how his sponsor is both safe and reliable despite giving no examples of how he determined these unexplained and unsubstantiated claims. Nearly every example of deceptive manipulation lodged against other channels is leveraged in Tom's own video and I find his eyebrow raising hypocrisy highly distracting. Tom could have taken a second crack at it with more problem solving and less pettiness but instead he played the role of victim when Veritasium responded.
 
Last edited:
Back
Top