• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Ed Self-Driving Cars: Pros, Cons, and Predictions

Evaluate Self-Driving Cars on a scale of 1-5 (1 = Terrible, 3 = Meh, 5 = Great)

  • 1

    Votes: 10 6.6%
  • 2

    Votes: 11 7.2%
  • 3

    Votes: 24 15.8%
  • 4

    Votes: 28 18.4%
  • 5

    Votes: 79 52.0%

  • Total voters
    152
  • Poll closed .
Technology-wize it is the same. The Teslas being used as robotaxis in Austin are just normal production cars with software tweeked for robotaxi use. When FSD v14 is released (any day now) it should have the same performance. The only reason Tesla owners aren't using their vehicles as robotaxis is risk aversion - ie. Ludditism. Tesla must play by the rules or they will be shut down, just like early motor cars that had to have a man with a red flag walk in front of the vehicle.

Red flag traffic laws
No, it's not. You have to start being critical of Tesla and Musk. They are getting better. But close is not enough.
 
I'm sorry, you are just wrong about this. Chance had very little to do with it. State of the art rechargeable battery technology for a hundred years has been heavy lead acid. We're talking 30kwh per KG. Not the 200 to 300 of a modern EV.
The improvements in battery chemistry didn't come about by chance. It also didn't improve because of a lack of interest. They had been trying to make a better battery for a century without much improvement.
This isn't quite right. Lead-acid batteries dominated because they were cheap, but there were better chemistries.

The Nickel-Cadmium battery was invented in 1899. Nicads have about 30% higher specific energy than Lead-acid, higher power and much higher cycle life. The practical specific energy is even higher because they can be fully discharged without damage, whereas the cycle life of Lead-acid dramatically reduces at high depth of discharge. They can be charged much faster too, typically in 1 hour vs 5 hours for lead-acid. They have been used in aircraft since the 1930s.

Consumer Nicads were phased out from 2006 because cadmium is highly toxic. This opened the door for another battery which was invented in 1967 to take over - Nickel Metal Hydride (NiMH). It has higher specific energy than Nicad, but lower power and shorter cycle life. Toyota used NiMH batteries in its Prius hybrid starting in 1997.

Rechargeable Lithium-ion batteries were invented in 1985, with commercial production beginning in 1991.

I started using Li-ion batteries in my electric powered RC model airplanes in 2003. Before that I used Nicads, then NiMH when they got powerful enough. The problem with Li-ion back then was that consumer batteries (mostly used in laptop computers and digital cameras) were optimized for high energy density rather than high power, which limited their use to lower powered models. Then some Chinese manufacturers started making Lithium-Polymer batteries for R/C model use. At first they were only rated for 6C (6 x times 1 hour discharge rate), but soon improved to 10C, then 25C, 40C etc. up to 60C and higher today (at 60C the battery can be completely discharged in just 1 minute!).

You may wonder why I am talking about this niche market. The reason is that this market demanded light weight high power batteries, encouraging manufacturers to develop them. Lead-acid dominated other markets where weight didn't matter, such as stationary storage and motor cars. You might notice however that devices such as laptops and cell phones didn't use Lead-acid even before Li-ion batteries existed. It's all about producing a product optimized for the application.

Because EVs weren't a thing since the 1930's, batteries were weren't developed for them. It wasn't that the technology inherently took so long to develop, but that there was little incentive to do so. Now that there is, big strides are being made. I can buy a Leaf HV battery repacked with modern cells that has over 3 times the capacity with only a small weight increase, or double the capacity while being lighter than the original! That's a huge improvement in a little over a decade. I'm sure there is more they can do too.

But it's not just the batteries that are being optimized. Tesla's Cybercab doesn't need a driver so it can be made smaller, suiting a smaller battery. It has inductive charging so it can recharge wherever it's parked, reducing battery capacity requirements even more. The car is made largely from plastic and glued together - 60% fewer parts, no welding, no painting - making it much cheaper to produce. That means Tesla can spend more money on the important part - the computer. The car will drive itself so well that many poeple will think twice about owning their own car when they can just hail a Cybercab whenever they need one for much less money. People will be shocked by how much their gas car is really costing them. We may see a whole new generation of people who don't own cars and have more money to spend on other stuff.
 
It is just plain stupid to make something that is supposed to carry people or cargo into something at is primarily carrying a computer - unless the goal is actually not at all about transport, and only about gathering data, and exploiting and selling that.
The computerized car is for the automotive industry what Monsanto Seeds are for agriculture: a way to keep customers completely dependent on you, selling them a car that they don't really own and can't use without the company's blessing and blind acceptance of their ToS.
 
No, it's not. You have to start being critical of Tesla and Musk. They are getting better. But close is not enough.
No, I'm not falling for that one. I've watched videos of people using FSD with zero interventions. I've read statements from people saying they use it all the time, and reviewers whom I trust. I've also read many comments from people who insist that it's anywhere from "not there yet" to a total fraud. Who am I to believe, some random bozo on the internet, or people who are actually using FSD? Of course you may say they are all fanboys paid by Musk Tesla to tell lies. If so, *plonk*.

"Close is not enough", is unscientific. When will it be good enough for you? Never, because 'close' is all we ever get with any technology. I'm not going to dismiss airliners because they still crash occasionally despite all their autopilot functions. I won't even dismiss Firefox because it crashes 2-3 times a day despite being updated a million times. The technology isn't perfect, but it's close enough.
 
It is just plain stupid to make something that is supposed to carry people or cargo into something at is primarily carrying a computer - unless the goal is actually not at all about transport, and only about gathering data, and exploiting and selling that.
The computerized car is for the automotive industry what Monsanto Seeds are for agriculture: a way to keep customers completely dependent on you, selling them a car that they don't really own and can't use without the company's blessing and blind acceptance of their ToS.
You know that all modern cars have a computer in them, right? The average car has 30 to 50 'Electronic Control Units', all talking to a central 'brain' that controls the various functions. What many don't have though is a well integrated system where every component talks the same langauge and is easily updatable to improve performance or add features. Tesla recently announced an improvement that deploys the airbags just before a crash, based on camera data. Meanwhile other autmakers often have to physically replace subsections that aren't working as well as they should due to software issues. That's one advantage of a 'software defined vehicle'.

When Tesla's are on the production line, the computer is one of the first things installed. Then it monitors the rest of the electronics and reports any anomalies while the car is being assembled. When the car reaches the end of the production line it drives itself to the carpark waiting for delivery. In the future it may drive directly to the customer or enter service as a robotaxi.

This isn't about locking people into a product they have to use, it's about giving them another choice. Some people love driving and would never give it up, others hate it but don't have a choice. When I took a job as a telephone exchange technician in 1976 I didn't have a car. This wasn't a problem because I lived close to where I worked. However it wasn't long before pressure was put on me to get a driver's license so I could install equipment at various locations by myself. So I took driving lessons and got my license. My first job involved a large piece of equipment almost too heavy for one person to handle, stuffed in the back of a Commer van which had horrible handling even without a top-heavy load - and me still learning how to drive. It's a miracle that I didn't crash it.

I look back on those times with horror. The things they expected me to do that would have OSH shut the whole place down if they tried it today. And the younger me so ignorant and trusting that I went along with it. I have no nostalgia for that time when I had to rely on the competence and integrity of humans.

Nor do I have any nostalgia for my first car, a Ford Cortina with column shift. I loved that vehicle - until I got a more modern car and realized what a horrible design the Cortina was. Then I had the same experience with a Nissan Sentra, which was a piece of ◊◊◊◊ compared to the Leaf. No doubt I would feel the same about the Leaf if I got a Tesla. The computerized car is nothing like Monsanto seeds, it's just a better car. Take out the steering wheel and pedals and have it drive where you tell it and it could be even better - like having your own personal chauffeur doing the stressful part. Every time I drive I wonder if this will be the day when I or someone else makes a mistake and the car gets smashed up or worse.

Everybody knows (or knew) someone who was involved in a bad accident. A friend of mine lost an eye and almsot died. My cousin was only a teenager when he was killed by a drunk driver. My brother's partner was in hospital for several weeks with a broken collar bone after someone pulled out of a driveway in front of her. Some other friends of mine were involved in a head-on when another driver fell asleep at the wheel. They were in hospital for several months. If computerized cars can save us from any of that then they are worth it, and you can stick your data exploitation conspiracy BS up your Monsanto.
 
Sure, my car has a computer in it but it is not connected to anything outside the car, and it does not need to be. I can guarantee that as long as nobody steals it or takes a crowbar to the doors, nobody will find out anything about anything. I can get information from it, engine codes, oil life and so forth, but that is trivial and private. It's possible that some information can be gotten out of it if I crash, but if I don't nobody (at least so far) is likely to read it if I don't.

I'm sorry Roger got lousy cars in the past. Some of us did not, and I still wish any car currently made could corner like my '68 Peugeot or my '72 Mercedes. I greatly appreciate some modern features like engine monitoring systems, and evolutionary developments like engines that consume no oil, but I like to drive a car, and trust myself out here in the country, where snow falls and animals cross the road, and frost heaves pop up overnight, and infrastruture is minimal, more than I would trust a self driving car, and I expect that will remain so for the foreseeable future.

Sure, it's possible that someday even those twisty dirt roads and such will have enough infrastructure, and the cars will have enough training, to allow them to negotiate safely on a snowy night, but from the purely utilitarian point of view it would be cheaper and more efficient, less disruptive to the environment, and probably safer, to set up a better bus system. If you just can't resist the technology of self driving vehicles, make a self driving bus that you can call to a reasonably close destination. If it made predictable stops within a reasonable distance, it could learn its route well without having to negotiate all the back roads. When the weather is good enough, you could go there by bicycle, or drive your old clunker in winter. Of course that won't happen, because although the gross cost would be lower, and the effect on the environment less, it would require public funding, and people who are willing to go into a lifetime of debt to buy the latest car boil over at the thought of paying a few bucks more in taxes, and the auto manufacturers are just fine with that.
 
I've probably said this before, but I'd like to see any autopilot system negotiate our single-track roads. Maybe some day, but it's going to take a while.
 
UK here, and the highway code says that a flash of headlights means only one thing - "I am here", i.e. alerting others to your presence.

Down my narrow, congested street it means "I am giving way to you", and if it didn't then driving on this road would be a nightmare with cars meeting and unable to progress without someone reversing. I wonder if robocars will be able to understand 'local rules' such as this?
 
UK here, and the highway code says that a flash of headlights means only one thing - "I am here", i.e. alerting others to your presence.

Down my narrow, congested street it means "I am giving way to you", and if it didn't then driving on this road would be a nightmare with cars meeting and unable to progress without someone reversing. I wonder if robocars will be able to understand 'local rules' such as this?
That is so strange as that's been in the HC forever, yet no one does that, as you say it's a signal to another driver that they can come through, why the HC hasn't been updated to reflect how every single driver uses a flash of their headlights seems perverse.
 
True dat.

Earlier this year I was on one of my car camping trips. These involve leaving the car switched on all night so I can sleep in it with the heating on. Snug as a bug in a rug. Just one teensy wee snag. The car has a software bug that keeps its sidelights on even when you activate the "lighting system off" switch. After various experiments with ways to cover the lights overnight, a bright spark (pun not really intended) on the MGEV forum discovered that pulling two fuses in the fuse box under the bonnet does the trick. There are even tweezers in the box to make this easier. Takes just over a minute, perfect solution (short of fixing the damn bug). The thing is, you have to remember to put them back before you move on, or you have no front lights.

So, I leave my camp site by the side of Loch Ness and proceed in the general direction of Fort Augustus, by way of the lochside Wade's road, which is single track much of the way. As I'm passing a long, unoccupied layby on my right, I see traffic trying to cram into a small passing place about 50 yards beyond this. I stop opposite the layby (where the road is effectively two-way) and attempt to flash my lights. Why is nobody moving on? Oops, I have no front lights. And this is actually illegal, because if your car has DRLs fitted you are not supposed to disable them.

Eventually the foremost of the oncoming vehicles realised, and moved forwards. The vehicle behind it was a cop car. Oops again, but fortunately he didn't put two and two together and realise that a car that should have been showing DRLs, and indeed flashing its lights, wasn't. I make a point of remembering to put the fuses back now.
 
I've probably said this before, but I'd like to see any autopilot system negotiate our single-track roads. Maybe some day, but it's going to take a while.
Or merge into a traffic jam that's moving at 50-60mph on a dual carriageway.

Will the next software update from Tesla include FSD with push'n'shove and the ability to pre-empt another driver is about to 'do' something based on an unusual speed and/or lane position? Will it know the highway code doesn't apply to drivers of Mercedes and BMWs?

Driving with humans requires theory of mind, not just the ability to navigate, stay in lane and collision avoidance.

There are just too many exceptions and corner cases that make FSD outside of simple environments a pipe dream. A skilled human will always be needed to be paying attention and ready to take over. However, FSD means that that human will probably be distracted by a TikTok video and/or not skilled enough to avert disaster when needed.

Like so much technology, FSD is a non-solution to a problem that doesn't exist.
 
Or merge into a traffic jam that's moving at 50-60mph on a dual carriageway.

Will the next software update from Tesla include FSD with push'n'shove and the ability to pre-empt another driver is about to 'do' something based on an unusual speed and/or lane position? Will it know the highway code doesn't apply to drivers of Mercedes and BMWs?

Driving with humans requires theory of mind, not just the ability to navigate, stay in lane and collision avoidance.

There are just too many exceptions and corner cases that make FSD outside of simple environments a pipe dream. A skilled human will always be needed to be paying attention and ready to take over. However, FSD means that that human will probably be distracted by a TikTok video and/or not skilled enough to avert disaster when needed.

Like so much technology, FSD is a non-solution to a problem that doesn't exist.
I wouldn't go that far. There are elderly and handicapped people that need mobility and want independence. I see the last mile from the transit station. I see light traffic busses not having a driver.

I just want it safe and affordable.
 
I wouldn't go that far. There are elderly and handicapped people that need mobility and want independence. I see the last mile from the transit station. I see light traffic busses not having a driver.

I just want it safe and affordable.
I think in areas where a bus route is fairly routine it ought to be possible to run driverless buses. It might take some time to educate people on how to interact with them, but it wasn't all that long ago that we were able to negotiate in many places with trams and trolleys that,while not owning the space they go in, have priority and a degree of lese majesté when in it.

Unfortunately, I imagine we will pay dearly the long-deferred cost of having ruined the existing mass transit infrastructure.
 
Or merge into a traffic jam that's moving at 50-60mph on a dual carriageway.

Will the next software update from Tesla include FSD with push'n'shove and the ability to pre-empt another driver is about to 'do' something based on an unusual speed and/or lane position? Will it know the highway code doesn't apply to drivers of Mercedes and BMWs?

Driving with humans requires theory of mind, not just the ability to navigate, stay in lane and collision avoidance.

There are just too many exceptions and corner cases that make FSD outside of simple environments a pipe dream. A skilled human will always be needed to be paying attention and ready to take over. However, FSD means that that human will probably be distracted by a TikTok video and/or not skilled enough to avert disaster when needed.

Like so much technology, FSD is a non-solution to a problem that doesn't exist.

Teslas in America seem to be able to do most of this already. And I think they could learn the single-track roads thing in time. It mostly hinges on negotiation with the oncoming driver as to who is going and who is giving way. Once that's mastered, an autopilot could probably manage the reversing part better than a human. (Yes I did scratch my beloved car coming down the Bealach na bà while being bullied by a big yellow sports car, why do you ask?)
 
Teslas in America seem to be able to do most of this already. And I think they could learn the single-track roads thing in time. It mostly hinges on negotiation with the oncoming driver as to who is going and who is giving way. Once that's mastered, an autopilot could probably manage the reversing part better than a human. (Yes I did scratch my beloved car coming down the Bealach na bà while being bullied by a big yellow sports car, why do you ask?)

I have v13.2.2 for FSD, often times there's a lane that's almost at a stop and a very fast moving lane next to it. Tesla changes lanes without asking so it proceeds to slowly merge into a car going 60mph while it's going 20mph. Just got out of an incident where it was nearly about to crash into the car on the incoming lane, had to swerve back into my lane and emergency brake not to hit the car in front.

I understand FSD requires full attention and it's my liability. But fr how does it want me to know when it's gonna change lanes, for me to check if that lane change is viable or not, then take over...

Sounds like there's a few issues...maybe in the next version of software!
 
Nobody is claiming that this technology is already at its maximum capability. And yet stats are already showing that Teslas under FSD are having significantly fewer accidents than human drivers. Obviously when something non-optimum happens it will be plastered all over the media, but the general performance is pretty impressive.
 
I wouldn't go that far. There are elderly and handicapped people that need mobility and want independence. I see the last mile from the transit station. I see light traffic busses not having a driver.

I just want it safe and affordable.
No chance. The software is going to be so complex that the people designing the system can't predict what it could do in the real world because there isn't enough time in the universe to cover all the test vectors. There will be a lot of "correct by design" used to handwave away intractable problems with verification and validation.
 
I think in areas where a bus route is fairly routine it ought to be possible to run driverless buses. It might take some time to educate people on how to interact with them, but it wasn't all that long ago that we were able to negotiate in many places with trams and trolleys that,while not owning the space they go in, have priority and a degree of lese majesté when in it.

Unfortunately, I imagine we will pay dearly the long-deferred cost of having ruined the existing mass transit infrastructure.
I live near a small rural town that is right off the Interstate. There is a bus route that travels basically about 50 miles South and then North on the Interstate. It pulls off the freeway at 5 exits, stops quickly at a bus stop and then gets back on the freeway. It is very useful, but not heavily used. So service is frequently cut and restored. It's needed by the poor but not much. Being able to replace the driver on this route should be about the easiest task that an autonomous vehicle could do.
 

Back
Top Bottom