self driving cars

[Moliere]

Like I said in the other thread. This isn't the future.

I agreed with you then and still agree. I think for self-driving cars to work, there will need to be an overhaul of the protocol for building and marking streets. I think a more robust and flexible public transit system is more likely to be the future in the cities, not individual cars that drive themselves. Although, I do think self-driving cars could more easily be implemented on the freeways, however it will take every car being adapted with the self-driving tools.

Of course, I'd just settle for something other than a lame bus system in Houston. What is it about Texas that makes people not want public transit?

[Omaha 680]

We are obviously a way off but it is definitely the future. If you want a robust and flexible public transit system in any American city outside of the Northeast corridor, self driving cars will have to be a big part of it. The future of self driving cars in America is not everyone having a Tesla in their garage. It's dedicated autonomous vehicle routes populated by public vehicles that can be flagged down like a Taxi.

[myboynoah]

I agreed with you then and still agree. I think for self-driving cars to work, there will need to be an overhaul of the protocol for building and marking streets. I think a more robust and flexible public transit system is more likely to be the future in the cities, not individual cars that drive themselves. Although, I do think self-driving cars could more easily be implemented on the freeways, however it will take every car being adapted with the self-driving tools.

Of course, I'd just settle for something other than a lame bus system in Houston. What is it about Texas that makes people not want public transit?

It will evolve. I don't think we are that far off from semi-autonomous (certainly on the highway). In town will require monitoring for a while until the technology advances. But eventually, fully autonomous vehicles will be an option, even in our lifetimes. The insurance industry will take a huge hit.

[Walter Sobchak]

It will evolve. I don't think we are that far off from semi-autonomous (certainly on the highway). In town will require monitoring for a while until the technology advances. But eventually, fully autonomous vehicles will be an option, even in our lifetimes. The insurance industry will take a huge hit.

Not gonna happen. If you look at the lines on the road in the video, it appears they weren't changed for the construction - the lines went right into the barrier; and the Tesla software obediently followed the markings right into the barrier.

Anyone that has spent any amount of time in the software industry writing code will immediately recognize the inherent problems with software-based autopilot systems. Every use case in such a system, even the most obscure and arcane scenario, would have to be addressed because of the severe life-threatening consequences. But typically, those types of extreme use cases would be ignored in a typical SDLC because they are simply too expensive (programmer-hours-wise) to address for the perceived benefit gained. Not so for an autopilot system, where the software has to be written far beyond any imaginable scenario for it to work "bug-free" every time.

But it won't be bug-free. Autonomous cars will hurt, maim, or kill their "beta testers" over the first few years of use in an operational "live" environment (rather than a controlled test environment) until the entire thing collapses under the weight of lawsuits brought by consumers harmed by autonomous vehicles. Software engineers might even need to carry malpractice insurance to shield them from bugs in their software algorithms that are determined to cause harm.

Don't get me wrong... I love my blind spot detection, lane drift warning sensors, 360° back up camera, smart headlamps, etc; I am not against all technologies in cars. But when it comes to driving the car I want to have control. As much as I'd like to let the RV drive itself so I can go back and make myself a sandwich... I've written enough code to know: there is no such thing as bug-free software.

[Moliere]

We are obviously a way off but it is definitely the future. If you want a robust and flexible public transit system in any American city outside of the Northeast corridor, self driving cars will have to be a big part of it. The future of self driving cars in America is not everyone having a Tesla in their garage. It's dedicated autonomous vehicle routes populated by public vehicles that can be flagged down like a Taxi.

Looks like we agree more than we disagree. Chances are the protocols for bigger streets will be adjusted (lane markings, intersections, etc.) and the self-driving cars will be restricted to those bigger streets. Neighborhood streets may still require a person to drive/steer/brake the car. That woulnd't be all that bad since I could just walk 400 yards to the main road and get a lift and it would significantly reduce the amount of traffic in my subdivision.

There will always be a need for some amount of human driving. I typically buy stuff from Lowes that requires a larger vehicle and I wonder if/how that type of thing would be incorporated. But mainly, I would just want an autonomous car to get to work adn back as it could save me significant time in commuting and allow me to take care of other tasks while "driving"

[All-American]

Not gonna happen. If you look at the lines on the road in the video, it appears they weren't changed for the construction - the lines went right into the barrier; and the Tesla software obediently followed the markings right into the barrier.

Anyone that has spent any amount of time in the software industry writing code will immediately recognize the inherent problems with software-based autopilot systems. Every use case in such a system, even the most obscure and arcane scenario, would have to be addressed because of the severe life-threatening consequences. But typically, those types of extreme use cases would be ignored in a typical SDLC because they are simply too expensive (programmer-hours-wise) to address for the perceived benefit gained. Not so for an autopilot system, where the software has to be written far beyond any imaginable scenario for it to work "bug-free" every time.

But it won't be bug-free. Autonomous cars will hurt, maim, or kill their "beta testers" over the first few years of use in an operational "live" environment (rather than a controlled test environment) until the entire thing collapses under the weight of lawsuits brought by consumers harmed by autonomous vehicles. Software engineers might even need to carry malpractice insurance to shield them from bugs in their software algorithms that are determined to cause harm.

Don't get me wrong... I love my blind spot detection, lane drift warning sensors, 360° back up camera, smart headlamps, etc; I am not against all technologies in cars. But when it comes to driving the car I want to have control. As much as I'd like to let the RV drive itself so I can go back and make myself a sandwich... I've written enough code to know: there is no such thing as bug-free software.

Software writers don't have to compete with perfection-- they have to compete with the status quo. The question is at what point will the technology be so advanced that driverless cars will be safer than human-operated cars. We might be there already.

[Pelado]

Software writers don't have to compete with perfection-- they have to compete with the status quo. The question is at what point will the technology be so advanced that driverless cars will be safer than human-operated cars. We might be there already.

It's possible that self-driving cars are or will be safer than the status quo. But when someone has an accident now, the presumption is that the cause of the accident was user error. When an accident happens in a self-driving car, wouldn't surprise me to see a lot more lawsuits against manufacturers and software engineers, even if the overall effect is fewer accidents.

Sent from my SM-G930V using Tapatalk

[Walter Sobchak]

Software writers don't have to compete with perfection-- they have to compete with the status quo.

You are wrong. With individual drivers, the liability is spread out among millions of drivers. With self-driving cars, the liability is concentrated upon the shoulders of the car manufacturers.

After a litany of wrongful death lawsuits, I doubt an automaker will be able to hide behind one of these:

avic-d1-warning.jpg

But I appreciate that you are dreamer. Keep dreaming.

[Joe Public]

Not gonna happen. If you look at the lines on the road in the video, it appears they weren't changed for the construction - the lines went right into the barrier; and the Tesla software obediently followed the markings right into the barrier.

Anyone that has spent any amount of time in the software industry writing code will immediately recognize the inherent problems with software-based autopilot systems. Every use case in such a system, even the most obscure and arcane scenario, would have to be addressed because of the severe life-threatening consequences. But typically, those types of extreme use cases would be ignored in a typical SDLC because they are simply too expensive (programmer-hours-wise) to address for the perceived benefit gained. Not so for an autopilot system, where the software has to be written far beyond any imaginable scenario for it to work "bug-free" every time.

But it won't be bug-free. Autonomous cars will hurt, maim, or kill their "beta testers" over the first few years of use in an operational "live" environment (rather than a controlled test environment) until the entire thing collapses under the weight of lawsuits brought by consumers harmed by autonomous vehicles. Software engineers might even need to carry malpractice insurance to shield them from bugs in their software algorithms that are determined to cause harm.

Don't get me wrong... I love my blind spot detection, lane drift warning sensors, 360° back up camera, smart headlamps, etc; I am not against all technologies in cars. But when it comes to driving the car I want to have control. As much as I'd like to let the RV drive itself so I can go back and make myself a sandwich... I've written enough code to know: there is no such thing as bug-free software.

I could see something along the lines of signal emitters that construction companies are required to use in the future that communicate a detour to self-driving cars that come into range.

[All-American]

You are wrong. With individual drivers, the liability is spread out among millions of drivers. With self-driving cars, the liability is concentrated upon the shoulders of the car manufacturers.

After a litany of wrongful death lawsuits, I doubt an automaker will be able to hide behind one of these:

[ATTACH=CONFIG]7802[/ATTACH]

But I appreciate that you are dreamer. Keep dreaming.

I take it that you are not quibbling with the idea that self-driving cars are (or ultimately will be) safer. It sounds like your point is one about allocation of costs, and whether a small group (manufacturers) should be willing to bear the risk of liability. If so, that has the makings of a classic Coasian bargaining issue: given the current allocation of rights, will transaction costs prevent the transition to the highest value use?

Call me a dreamer, then, because I think the transition is absolutely possible and even desirable.

[Shaka]

Not gonna happen. If you look at the lines on the road in the video, it appears they weren't changed for the construction - the lines went right into the barrier; and the Tesla software obediently followed the markings right into the barrier.

Anyone that has spent any amount of time in the software industry writing code will immediately recognize the inherent problems with software-based autopilot systems. Every use case in such a system, even the most obscure and arcane scenario, would have to be addressed because of the severe life-threatening consequences. But typically, those types of extreme use cases would be ignored in a typical SDLC because they are simply too expensive (programmer-hours-wise) to address for the perceived benefit gained. Not so for an autopilot system, where the software has to be written far beyond any imaginable scenario for it to work "bug-free" every time.

But it won't be bug-free. Autonomous cars will hurt, maim, or kill their "beta testers" over the first few years of use in an operational "live" environment (rather than a controlled test environment) until the entire thing collapses under the weight of lawsuits brought by consumers harmed by autonomous vehicles. Software engineers might even need to carry malpractice insurance to shield them from bugs in their software algorithms that are determined to cause harm.

Don't get me wrong... I love my blind spot detection, lane drift warning sensors, 360° back up camera, smart headlamps, etc; I am not against all technologies in cars. But when it comes to driving the car I want to have control. As much as I'd like to let the RV drive itself so I can go back and make myself a sandwich... I've written enough code to know: there is no such thing as bug-free software.

It's not nearly as hard or as dire as you predict. This type of automation already exists on a smaller scale in many of the fully automated factories around the world where all kinds of self-drive material handling vehicles are running around a factory delivering parts and supplies. It looks like utter chaos if you get a chance to view such a factory from above but it really works. The automakers are adapting this on a larger scale with the added benefit of GPS.

[Jeff Lebowski]

So what happens in canyons or tunnels where you lost GPS connection?

[Shaka]

So what happens in canyons or tunnels where you lost GPS connection?

Same thing that happens in a factory with a roof. Sensors in the vehicle and tunnel talk to each other.

[Jeff Lebowski]

Same thing that happens in a factory with a roof. Sensors in the vehicle and tunnel talk to each other.

Sure, but I'm thinking more short term. I imagine it would take a long time to get all of the highways set up with sensors.

[Walter Sobchak]

It sounds like your point is one about allocation of costs, and whether a small group (manufacturers) should be willing to bear the risk of liability. If so, that has the makings of a classic Coasian bargaining issue: given the current allocation of rights, will transaction costs prevent the transition to the highest value use?

You've probably seen this (first minute is adequate):

I'm reminded of a time when my son, about 8 years old, asked if he could ride his bike to a friend's house. The distance to the friend's home was fairly short (by my childhood bike-riding standards) and would take about 8-10 minutes. I said, sure, see ya later. Be home by dinner or call. Now, let me state, that we are (in all likelihood) the only parents in the city limits that do not make our kids wear bike helmets. I know... call child protective services. Anyway... after my son left, about 10 minutes passed and then the doorbell rang. One of the neighbors, bless her heart, had picked up my son and brought him home because she saw him biking without his helmet and decided that she must quickly intervene. Crisis averted!

As such, I have one quibble with the video above... one dead child may be worth the least to a personal injury lawyer, but one dead child killed by an autonomous vehicle will rally hundreds of thousands, nay millions, of mothers against the evil corporations responsible for putting their children in harms way. When that happens... game over.