In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
This would be fundamentally unsafe. Something as massive and dangerous as a car cannot afford to trust anything external. If autonomous vehicles trust external guidance information, people would absolutely figure out ways to mess with them.
This will happen, just as we already have drone swarm demos and other machines that work in unison. This will happen because it adds safety and efficiency. This does not mean it will be the only metric by which this machines operate, but will an additional way by which they can stay safe and be more efficient.
Imagine there is a highway of fully autonomous vehicles driving down a congested road. A deer runs across which causes the first car to apply the breaks. The car right behind it uses its cameras and other sensors to detect that the car has slowed down so it slows down—and with a reaction time much faster than a human under the best circumstances—which causes the car behind it to then slow down when its camera and sensors noticing the change. This happens all the way back and causes a traffic wave, which we're all familiar with even if most of us didn't even know it had a name.
If fully autonomous automobiles also could be connected in other ways, like being able to pass on information to the car behind it, the car could start reacting to a potential event even sooner, which could help mitigate the potential for even more accidents, which cause property damage, injury, and hurt efficiency a lot more if the only data that is had is from the object directly in your view. To imagine the difference, think of how haplessly and inefficiently people file an area without direction v soldiers in formation being given commands on when to step. Those soldiers still have free will and don't have to listen to the data being received upstream, especially if they violate safety, like if a marching solider in from of them collapses (like if a car's electric engine died), but they can use this info to be faster and more efficient without bumping into each other as often.
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
This would be fundamentally unsafe. Something as massive and dangerous as a car cannot afford to trust anything external. If autonomous vehicles trust external guidance information, people would absolutely figure out ways to mess with them.
This will happen, just as we already have drone swarm demos and other machines that work in unison. This will happen because it adds safety and efficiency. This does not mean it will be the only metric by which this machines operate, but will an additional way by which they can stay safe and be more efficient.
Imagine there is a highway of fully autonomous vehicles driving down a congested road. A deer runs across which causes the first car to apply the breaks. The car right behind it uses its cameras and other sensors to detect that the car has slowed down so it slows down—and with a reaction time much faster than a human under the best circumstances—which causes the car behind it to then slow down when its camera and sensors noticing the change. This happens all the way back and causes a traffic wave, which we're all familiar with even if most of us didn't even know it had a name.
If fully autonomous automobiles also could be connected in other ways, like being able to pass on information to the car behind it, the car could start reacting to a potential event even sooner, which could help mitigate the potential for even more accidents, which cause property damage, injury, and hurt efficiency a lot more if the only data that is had is from the object directly in your view. To imagine the difference, think of how haplessly and inefficiently people file an area without direction v soldiers in formation being given commands on when to step. Those soldiers still have free will and don't have to listen to the data being received upstream, especially if they violate safety, like if a marching solider in from of them collapses (like if a car's electric engine died), but they can use this info to be faster and more efficient without bumping into each other as often.
Drone swarms are not life-critical.
They are also generally operated in the hundreds-to-thousands and with very limited interoperability concerns (i.e., swarms are generally composed of identical devices, not different models or different manufacturers). To mess with your drone swarm, somebody would need to get one of your drones or spend a lot of time reverse engineering the protocol practically from first principles.
For cars, we're talking tens of millions, all of which would need to talk substantially the same swarm coordination protocol. To mess with it, someone would just need to pull the computer from one car—not necessarily the same make as yours—and give that computer false input from its sensors. Much broader availability of parts leads to a much simpler problem space for someone trying to break the system.
The entire point of swarming protocols like you're discussing is to let an individual member of the swarm operate closer to—or even outside—uncoordinated safety limits because it can be sure the other members are all doing the same. You can run at higher speed than your sensors can adequately scan for obstacles in time for your brakes to react, because you trust the scan from the car ahead of you.
Salvage a computer from a swarming car and configure it to lie to others, and you could do some really horrific stuff. Pick a non-swarming car and convince all the cars around it that it's swarm-capable, then tell them all there's a baby in the street and they need to engage panic-braking. Tell a swarm at high speed on a highway that the road doesn't go this way, it goes that way, and have them all leave the highway at a hundred miles an hour into a ditch. Worse, any failure of the system inherently affects many people, because swarming behavior only matters for groups.
Sure, swarming sounds neat. Doesn't mean it's possible to do safely.
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
This would be fundamentally unsafe. Something as massive and dangerous as a car cannot afford to trust anything external. If autonomous vehicles trust external guidance information, people would absolutely figure out ways to mess with them.
This will happen, just as we already have drone swarm demos and other machines that work in unison. This will happen because it adds safety and efficiency. This does not mean it will be the only metric by which this machines operate, but will an additional way by which they can stay safe and be more efficient.
Imagine there is a highway of fully autonomous vehicles driving down a congested road. A deer runs across which causes the first car to apply the breaks. The car right behind it uses its cameras and other sensors to detect that the car has slowed down so it slows down—and with a reaction time much faster than a human under the best circumstances—which causes the car behind it to then slow down when its camera and sensors noticing the change. This happens all the way back and causes a traffic wave, which we're all familiar with even if most of us didn't even know it had a name.
If fully autonomous automobiles also could be connected in other ways, like being able to pass on information to the car behind it, the car could start reacting to a potential event even sooner, which could help mitigate the potential for even more accidents, which cause property damage, injury, and hurt efficiency a lot more if the only data that is had is from the object directly in your view. To imagine the difference, think of how haplessly and inefficiently people file an area without direction v soldiers in formation being given commands on when to step. Those soldiers still have free will and don't have to listen to the data being received upstream, especially if they violate safety, like if a marching solider in from of them collapses (like if a car's electric engine died), but they can use this info to be faster and more efficient without bumping into each other as often.
Drone swarms are not life-critical.
They are also generally operated in the hundreds-to-thousands and with very limited interoperability concerns (i.e., swarms are generally composed of identical devices, not different models or different manufacturers). To mess with your drone swarm, somebody would need to get one of your drones or spend a lot of time reverse engineering the protocol practically from first principles.
For cars, we're talking tens of millions, all of which would need to talk substantially the same swarm coordination protocol. To mess with it, someone would just need to pull the computer from one car—not necessarily the same make as yours—and give that computer false input from its sensors. Much broader availability of parts leads to a much simpler problem space for someone trying to break the system.
The entire point of swarming protocols like you're discussing is to let an individual member of the swarm operate closer to—or even outside—uncoordinated safety limits because it can be sure the other members are all doing the same. You can run at higher speed than your sensors can adequately scan for obstacles in time for your brakes to react, because you trust the scan from the car ahead of you.
Salvage a computer from a swarming car and configure it to lie to others, and you could do some really horrific stuff. Pick a non-swarming car and convince all the cars around it that it's swarm-capable, then tell them all there's a baby in the street and they need to engage panic-braking. Tell a swarm at high speed on a highway that the road doesn't go this way, it goes that way, and have them all leave the highway at a hundred miles an hour into a ditch. Worse, any failure of the system inherently affects many people, because swarming behavior only matters for groups.
Sure, swarming sounds neat. Doesn't mean it's possible to do safely.
It's folly to take one example of how devices can talk to each other and then assume it would be done exactly there same way. It won't! Those drones were programmed together to a single unit, which autos, each with different destinations won't.
No, it doesn't take a single act of terror to mess up the whole system. That's like taking down one street light would affect all them equally because they're all connected by roads. All the other systems will still be in place, as well as other vehicles talking to other vehicles, as well as GPS and tower data giving automobiles a much better picture of what is going on with the road than what you have at any given time as the driver. Autonomous driving saves lives!
Your concept reminds me of how biometrics are so easily thwarted in movies. Take Minority Report, for example. Tom Cruise has his eyes replaced, but saves the old ones for the biometric scan to get back into the gov't building that once employed him before he was a fugitive. His access would've absolutely been revoked. I've known people that knew they were getting fired before they were called in for an exit interview because their logins and keycards had stopped working.
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
This would be fundamentally unsafe. Something as massive and dangerous as a car cannot afford to trust anything external. If autonomous vehicles trust external guidance information, people would absolutely figure out ways to mess with them.
As for the idea of integrating LiDAR data, of course they will.
In the 90s, self-driving cars used cameras for lateral motion and position estimation. Cameras have high angular resolution, so they're great for determining which lane a car in the camera's view is in. None of the features of cars (size, taillight spacing, &c.) are standardized, though, so it's hard to tell if you're looking at a small car which is close, or a large car which is further away.
Around 1993, they started adding RADAR for longitudinal position and motion estimation. These sensors had great distance and speed resolution, but poor selectivity. They work in a pretty broad cone, so it's hard to tell if the distance and motion you get back are from the car in front of you, or a car a lane (or even two!) to the side. The data from both sensors was integrated and gave a pretty solid combined estimate of the world outside the car.
LiDAR gives good angular resolution and good distance resolution. Further, it's a lot less susceptible to external interference. It can't tell which light at an intersection is illuminated, though. I expect LiDAR to be used for most guidance with a color camera (or an array of several) used to identify signs and lights, and as backup in case the LiDAR returns nonsensical data.
Data from several internal sensors will probably be used as well. GPS receiver, at least one motion-compensated level, and at least two chassis IMUs (which might be used to synthesize the level). I would actually expect IMUs in each hub (along with the wheel speed sensor and a mic) to monitor flex and control semi-active suspension components.
Tesla is able to estimate distance and vehicle size with a high level of accuracy using just cameras. Karpathy has some great demos of this.
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
No it's not. It's already here. What you're confusing is some future, absolute concept of self-driving from sci-fi at some imaginary point in time, and how technology evolves and what "self" actually means. We've been moving toward this since the invention of the automobile. There is no all or nothing. These are small incremental steps that lead to more safety and efficiency every single year.
Let me know when I can buy a car, get in it, and tell it to take me to Walmart. Not gonna happen any time soon. in this age of pandemics no one would dare to ride in some autonomous taxi that god-knows-who just vomited in.
Why would you buy a car that someone just vomited in?
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
This would be fundamentally unsafe. Something as massive and dangerous as a car cannot afford to trust anything external. If autonomous vehicles trust external guidance information, people would absolutely figure out ways to mess with them.
As for the idea of integrating LiDAR data, of course they will.
In the 90s, self-driving cars used cameras for lateral motion and position estimation. Cameras have high angular resolution, so they're great for determining which lane a car in the camera's view is in. None of the features of cars (size, taillight spacing, &c.) are standardized, though, so it's hard to tell if you're looking at a small car which is close, or a large car which is further away.
Around 1993, they started adding RADAR for longitudinal position and motion estimation. These sensors had great distance and speed resolution, but poor selectivity. They work in a pretty broad cone, so it's hard to tell if the distance and motion you get back are from the car in front of you, or a car a lane (or even two!) to the side. The data from both sensors was integrated and gave a pretty solid combined estimate of the world outside the car.
LiDAR gives good angular resolution and good distance resolution. Further, it's a lot less susceptible to external interference. It can't tell which light at an intersection is illuminated, though. I expect LiDAR to be used for most guidance with a color camera (or an array of several) used to identify signs and lights, and as backup in case the LiDAR returns nonsensical data.
Data from several internal sensors will probably be used as well. GPS receiver, at least one motion-compensated level, and at least two chassis IMUs (which might be used to synthesize the level). I would actually expect IMUs in each hub (along with the wheel speed sensor and a mic) to monitor flex and control semi-active suspension components.
Tesla is able to estimate distance and vehicle size with a high level of accuracy using just cameras. Karpathy has some great demos of this.
Doubtful Apple is building a car. But in the event they are be assured that its many years away
You must've just gotten out of a coma if you missed the last 5 and a half years of rumors and three years of patents. Also, Donald Trump is president and there's a global pandemic, sorry!
In my opinion, until the human driver is completely eliminated from the equation the autonomous vehicle is a pipe dream. I watched a PBS documentary about the status of the technology. The really interesting part concerned interviews with the test vehicle drivers who reported that the super cautious nature of the software caused road rage in human drivers. They also reported that some drivers (aka assholes) intentionally tried to run them off the road or get the vehicle to crash. Humans are emotional, irrational, evil by nature. And as I have said before no self-driving car would have a chance against an 80 year old blue hair diving her 1975 Buick Electra gunboat.
The only way self driving cars will ever be a thing is if they are connected to every other car around them so they can talk to each other about changing lanes, etc. It will be a long time before all of the cars or at least 80-90% are replaced by cars that can do this, so self driving is still pretty far off despite what people want us to believe.
No it's not. It's already here. What you're confusing is some future, absolute concept of self-driving from sci-fi at some imaginary point in time, and how technology evolves and what "self" actually means. We've been moving toward this since the invention of the automobile. There is no all or nothing. These are small incremental steps that lead to more safety and efficiency every single year.
Let me know when I can buy a car, get in it, and tell it to take me to Walmart. Not gonna happen any time soon. in this age of pandemics no one would dare to ride in some autonomous taxi that god-knows-who just vomited in.
Why would you buy a car that someone just vomited in?
I can only think of 2 reasons you'd have a car that someone just vomited in:-
- Family. You have teenage kids or adults getting revenge on said kids using the chemical weapon known as playdate at grandmas house.
- Car share but think Car share company would have access to environmental sensors in the car to know it needs to leave the road for service and add a fee to the last users bill.
Indeed car share cars could have all sorts of systems build in to keep people moving in times like this and plenty of time between customers with no one in the care to deploy a disinfectant mist, UV lights then high-grade filters to be clean and ready. Plus they need to top up the tank battery wise a couple of times a day so distributed detailing crews could keep them cleaner than clean as they drop by the charge points.
Still, I think Apple will be aiming at the Family / Company extension to the home or office self-driving car market. More than car-share market.
Comments
Imagine there is a highway of fully autonomous vehicles driving down a congested road. A deer runs across which causes the first car to apply the breaks. The car right behind it uses its cameras and other sensors to detect that the car has slowed down so it slows down—and with a reaction time much faster than a human under the best circumstances—which causes the car behind it to then slow down when its camera and sensors noticing the change. This happens all the way back and causes a traffic wave, which we're all familiar with even if most of us didn't even know it had a name.
If fully autonomous automobiles also could be connected in other ways, like being able to pass on information to the car behind it, the car could start reacting to a potential event even sooner, which could help mitigate the potential for even more accidents, which cause property damage, injury, and hurt efficiency a lot more if the only data that is had is from the object directly in your view. To imagine the difference, think of how haplessly and inefficiently people file an area without direction v soldiers in formation being given commands on when to step. Those soldiers still have free will and don't have to listen to the data being received upstream, especially if they violate safety, like if a marching solider in from of them collapses (like if a car's electric engine died), but they can use this info to be faster and more efficient without bumping into each other as often.
They are also generally operated in the hundreds-to-thousands and with very limited interoperability concerns (i.e., swarms are generally composed of identical devices, not different models or different manufacturers). To mess with your drone swarm, somebody would need to get one of your drones or spend a lot of time reverse engineering the protocol practically from first principles.
For cars, we're talking tens of millions, all of which would need to talk substantially the same swarm coordination protocol. To mess with it, someone would just need to pull the computer from one car—not necessarily the same make as yours—and give that computer false input from its sensors. Much broader availability of parts leads to a much simpler problem space for someone trying to break the system.
The entire point of swarming protocols like you're discussing is to let an individual member of the swarm operate closer to—or even outside—uncoordinated safety limits because it can be sure the other members are all doing the same. You can run at higher speed than your sensors can adequately scan for obstacles in time for your brakes to react, because you trust the scan from the car ahead of you.
Salvage a computer from a swarming car and configure it to lie to others, and you could do some really horrific stuff. Pick a non-swarming car and convince all the cars around it that it's swarm-capable, then tell them all there's a baby in the street and they need to engage panic-braking. Tell a swarm at high speed on a highway that the road doesn't go this way, it goes that way, and have them all leave the highway at a hundred miles an hour into a ditch. Worse, any failure of the system inherently affects many people, because swarming behavior only matters for groups.
Sure, swarming sounds neat. Doesn't mean it's possible to do safely.
No, it doesn't take a single act of terror to mess up the whole system. That's like taking down one street light would affect all them equally because they're all connected by roads. All the other systems will still be in place, as well as other vehicles talking to other vehicles, as well as GPS and tower data giving automobiles a much better picture of what is going on with the road than what you have at any given time as the driver. Autonomous driving saves lives!
Your concept reminds me of how biometrics are so easily thwarted in movies. Take Minority Report, for example. Tom Cruise has his eyes replaced, but saves the old ones for the biometric scan to get back into the gov't building that once employed him before he was a fugitive. His access would've absolutely been revoked. I've known people that knew they were getting fired before they were called in for an exit interview because their logins and keycards had stopped working.
https://www.engadget.com/2016-12-28-tesla-autopilot-predicts-crash.html
The sooner we take human error/stupidity out of the equation the better.
https://en.wikipedia.org/wiki/Apple_electric_car_project
https://www.patentlyapple.com/patently-apple/autonomous-vehicle-technology/