A human as the final arbiter is a reasonable criteria especially for testing. That's a difference as I understand it between Boeing and Airbus design philosophy and it's Airbus' aircraft that crash when their computers get boggled by iced over pitot tubes: Air France being the latest.
Autonomous is likely a ways down the timeline: but augmentation through advanced tech is already appearing and will continue to add safety to vehicle performance in the near future (one minor one, we're still using naked eyeballs at night? I want an IR display at the bottom of my windshield in a "heads up display" implementation showing me the FULL road ahead at night... ). LONG before anything like a fully autonomous vehicle even appears.
Door locking mechanisms have been unlocked for a couple of decades, and have nothing to do with the type of actual hacking we're being told about today.
I don't know this Car & Driver Blog seems to disagree.
In particular, the Miller/Valasek Jeep Cherokee wireless hack this past Summer, seems to prove a hack of this nature which takes over the entire vehical is indeed possible.
EDIT: just read the linked Bloomberg article ... why are you focusing on the door locking mechanisms? They took over and shut the entire car down wirelessly ... and whether or not it has been successfully accomplished before now, I'm not sure anyone can trust that such a thing won't be possible in the future. The only way to prevent it is to put a physical barrier between the CAN bus and the Infotainment system. And autonomous vehicle manufactures don't seem to be interested in doing that, since piloting the car without a driver requires input from the information gleaned from the web. So it's a concern that must be addressed.
That article agrees with what I've been saying. I don't know why you bothered to link to it. The closest it comes to what you're saying is that there COULD be a problem in the future. Coulda, woulda, shoulda.
i mentioned door locks, because I was responding to a post by beltsbear that mentioned door locks. Didn't you actually read the whole post?
The reality is that no car has ever been hacked. That is, the only cars that have been hacked have been owned by the researchers doing the hacking. No other car has ever been hacked. The researchers have had to get into the electronics physically to hack it. In one of the two cases, they needed a laptop connected to the diagnostic port when the car was in use in order to control it. This is another situation where people are being frightened by something that hasn't happened in reality, and is very unlikely to happen.
There have been a lot of problems with these cars. Google has software engineers riding in every car. When odd situations come up, and I'll name a couple, they can get out of trouble right away, and even do some on the fly reprogramming.
A problem was a guy standing on a corner wearing a t-shirt saying STOP. The various sensors in the car couldn't distinguish whether this was a real stop sign or not, and refused to go.
Another problem was with another guy on a bicycle at a corner. As many bikers do, he was moving back and forth a bit on his bike waiting for the light to change. The car had the right of way, but just moved forward an inch when the bike moved backwards, and stopped when it moved forwards. The car went nowhere.
There have been many problems like these two. They didn't result in accidents, but they did result in the cars going nowhere, and holding up traffic.
They are very rarely published (because we all want to enjoy the illusion that self-driving cars are "just around the corner"), but there are many, many, spectacular fails involved with testing self-driving cars. Despite all the ink spent on self-driving cars almost no one bothers to mention that they are a long way from being ready for prime time and that the whole venture is likely even impossible from the start.
Currently, self-driving cars simply don't work. Period. That is at least if by "working" it is meant that they can drive themselves in the same way and with the same level of safety and accuracy that the average half-drunk high-school student is able to. A simple downpour (heavy rain) completely stymies most self-driving cars. As does darkness, fog, etc. The sensors simply can't keep up and they can't tell where they are in such conditions, or where the road is. A car that only works on a sunny day on a clear highway in California might be exciting to California residents, but not to anyone else.
Not only that, but the more one looks into self-driving cars, the more obvious it becomes that it's essentially a dead end technology, because the problems that need to be overcome are almost completely intractable.
The easiest (and possibly the only) solution to the problems self-driving cars face, is simply embedding a strip of metal in the middle of the road for the car to "follow," so that's likely what will eventually be done. It's cheap and it solves all the problems. But even though we might still refer to them as such, they won't be "self-driving" cars after that will they?
I don't think it's a dead end technology. But I do think it will take time. If somehow, driverless cars were mandated in two years, and that every manual car had to come off the road, it would already work well.
i believe that what's going to happen is that we'll continue to see more more semi autonomous features coming out, such as the avoidance technologies some companies already have on the road. This will sneak up on us rather than smacking us on the head.
There have been many problems like these two. They didn't result in accidents, but they did result in the cars going nowhere, and holding up traffic.
They are very rarely published (because we all want to enjoy the illusion that self-driving cars are "just around the corner"), but there are many, many, spectacular fails involved with testing self-driving cars. Despite all the ink spent on self-driving cars almost no one bothers to mention that they are a long way from being ready for prime time and that the whole venture is likely even impossible from the start.
Currently, self-driving cars simply don't work. Period. That is at least if by "working" it is meant that they can drive themselves in the same way and with the same level of safety and accuracy that the average half-drunk high-school student is able to. A simple downpour (heavy rain) completely stymies most self-driving cars. As does darkness, fog, etc. The sensors simply can't keep up and they can't tell where they are in such conditions, or where the road is. A car that only works on a sunny day on a clear highway in California might be exciting to California residents, but not to anyone else.
Not only that, but the more one looks into self-driving cars, the more obvious it becomes that it's essentially a dead end technology, because the problems that need to be overcome are almost completely intractable.
The easiest (and possibly the only) solution to the problems self-driving cars face, is simply embedding a strip of metal in the middle of the road for the car to "follow," so that's likely what will eventually be done. It's cheap and it solves all the problems. But even though we might still refer to them as such, they won't be "self-driving" cars after that will they?
They are very rarely published (because we all want to enjoy the illusion that self-driving cars are "just around the corner"), but there are many, many, spectacular fails involved with testing self-driving cars. Despite all the ink spent on self-driving cars almost no one bothers to mention that they are a long way from being ready for prime time and that the whole venture is likely even impossible from the start.
Currently, self-driving cars simply don't work. Period. That is at least if by "working" it is meant that they can drive themselves in the same way and with the same level of safety and accuracy that the average half-drunk high-school student is able to. A simple downpour (heavy rain) completely stymies most self-driving cars. As does darkness, fog, etc. The sensors simply can't keep up and they can't tell where they are in such conditions, or where the road is. A car that only works on a sunny day on a clear highway in California might be exciting to California residents, but not to anyone else.
Not only that, but the more one looks into self-driving cars, the more obvious it becomes that it's essentially a dead end technology, because the problems that need to be overcome are almost completely intractable.
The easiest (and possibly the only) solution to the problems self-driving cars face, is simply embedding a strip of metal in the middle of the road for the car to "follow," so that's likely what will eventually be done. It's cheap and it solves all the problems. But even though we might still refer to them as such, they won't be "self-driving" cars after that will they?
I don't think it's a dead end technology. But I do think it will take time. If somehow, driverless cars were mandated in two years, and that every manual car had to come off the road, it would already work well.
i believe that what's going to happen is that we'll continue to see more more semi autonomous features coming out, such as the avoidance technologies some companies already have on the road. This will sneak up on us rather than smacking us on the head.
I read just fine. Both linked articles very clearly state a car has been wirelessly hacked and high jacked remotely. Period. It's been done, it will be done again.
youre assertion that it's never been done and there is no cause for concern is apparently incorrect.
Self-driving cars generally won't commit traffic infractions, can't sign traffic tickets, and don't do wireless fund transfers to pay tickets. So I'm not surprised that legislators are eager to pass laws making self-driving cars less practical and desirable. You've got places like Ferguson, MO where the second largest source of revenue for the local government was tickets and associated fees (like "court costs"). When Ferguson court revenues exceeded $2 million in 2012, the city
manager sent the police chief an internal email that said "Awesome!
Thanks!" You think they are going to give up that kind of revenue without a fight?
The reality is that no car has ever been hacked. That is, the only cars that have been hacked have been owned by the researchers doing the hacking. No other car has ever been hacked. The researchers have had to get into the electronics physically to hack it. In one of the two cases, they needed a laptop connected to the diagnostic port when the car was in use in order to control it. This is another situation where people are being frightened by something that hasn't happened in reality, and is very unlikely to happen.
Of course cars have been hacked. This is way beyond research as well. In the article below Jeeps were hacked with NO PHYSICAL access via the airwaves. Doors have been unlocked and this has been used in actual cases of theft.
I don't think it's a dead end technology. But I do think it will take time. If somehow, driverless cars were mandated in two years, and that every manual car had to come off the road, it would already work well.
i believe that what's going to happen is that we'll continue to see more more semi autonomous features coming out, such as the avoidance technologies some companies already have on the road. This will sneak up on us rather than smacking us on the head.
I read just fine. Both linked articles very clearly state a car has been wirelessly hacked and high jacked remotely. Period. It's been done, it will be done again.
youre assertion that it's never been done and there is no cause for concern is apparently incorrect.
I think it's inevitable that driverless cars will be the norm. I think California is just thinking that these cars will run amok and start running over pedestrians. Surely, the self-drive folks have implemented safeguards to prevent such a thing?
This is all during testing. I think it's perfectly reasonable to mandate a human driver when you're testing new self-driving software. All code has bugs. There must be a safety-fallback of some kind when bugs have the potential to harm human lives - especially when the tests are in a location open to the public.
I would assume that in the future, after the technology has stabilized and been proven safe, that rules like this will (eventually) be removed.
That may not be for 20 years or more (if ever), but that's irrelevant. I think it is perfectly valid to require that there be a driver (and that he be held liable for the car's actions) at this time and until the tech has truly become mature enough to prove that such laws are no longer necessary.
I personally can't wait for that to happen, but there's no way I'm going to risk my life by having people go driverless before all the problems have been solved.
I don't think it's a dead end technology. But I do think it will take time. If somehow, driverless cars were mandated in two years, and that every manual car had to come off the road, it would already work well.
I don't know about that. I think there's more work that needs to be done, even if other cars are removed from the road. But I do agree that a completely controlled environment would be easier for the software to manage.
Comments
Autonomous is likely a ways down the timeline: but augmentation through advanced tech is already appearing and will continue to add safety to vehicle performance in the near future (one minor one, we're still using naked eyeballs at night? I want an IR display at the bottom of my windshield in a "heads up display" implementation showing me the FULL road ahead at night... ). LONG before anything like a fully autonomous vehicle even appears.
i mentioned door locks, because I was responding to a post by beltsbear that mentioned door locks. Didn't you actually read the whole post?
I don't think it's a dead end technology. But I do think it will take time. If somehow, driverless cars were mandated in two years, and that every manual car had to come off the road, it would already work well.
i believe that what's going to happen is that we'll continue to see more more semi autonomous features coming out, such as the avoidance technologies some companies already have on the road. This will sneak up on us rather than smacking us on the head.
https://static.googleusercontent.com/media/www.google.com/en//selfdrivingcar/files/reports/report-0515.pdf
youre assertion that it's never been done and there is no cause for concern is apparently incorrect.
I would assume that in the future, after the technology has stabilized and been proven safe, that rules like this will (eventually) be removed.
That may not be for 20 years or more (if ever), but that's irrelevant. I think it is perfectly valid to require that there be a driver (and that he be held liable for the car's actions) at this time and until the tech has truly become mature enough to prove that such laws are no longer necessary.
I personally can't wait for that to happen, but there's no way I'm going to risk my life by having people go driverless before all the problems have been solved.
I don't know about that. I think there's more work that needs to be done, even if other cars are removed from the road. But I do agree that a completely controlled environment would be easier for the software to manage.