Future Cars: Electric and Autonomous Vehicles

Roads - Rails - Sidewalks - Bikeways
moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 18th, 2015, 1:59 pm

Also if we want to remove traffic from the roads and remove congestion what we should be doing is incentivising all businesses that can to have employees work from home where it is applicable. That is something that doesn't take building any new lanes.

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 18th, 2015, 2:16 pm

But cars that are capable of networking with one and other can handle managing the network MUCH better. When all 250 cars ahead of you in the 1/4 mile before you have to exit understand that your vehicle needs to be in line to exit the freeway then it becomes a math problem to get your car where it needs to be in the most efficient manner. A pretty simple math problem.
Ha!

You are seriously underestimating the complexity of implementing distributed systems.

You have what to me feels like an NP-complete problem tackled by a distributed computing system that has a high probability of node failure.

Right now we have a hard enough time just keeping weather forecasting up and running.

I don't dispute that some communication among vehicles can make things better but people who think that will somehow produce an oracle capable of making perfect decisions every time are deluding themselves.

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 18th, 2015, 3:49 pm

But cars that are capable of networking with one and other can handle managing the network MUCH better. When all 250 cars ahead of you in the 1/4 mile before you have to exit understand that your vehicle needs to be in line to exit the freeway then it becomes a math problem to get your car where it needs to be in the most efficient manner. A pretty simple math problem.
Ha!

You are seriously underestimating the complexity of implementing distributed systems.

You have what to me feels like an NP-complete problem tackled by a distributed computing system that has a high probability of node failure.

Right now we have a hard enough time just keeping weather forecasting up and running.

I don't dispute that some communication among vehicles can make things better but people who think that will somehow produce an oracle capable of making perfect decisions every time are deluding themselves.
Weather patterns are much less predictable than cars traveling down lanes to a known location. It isn't as complex as you make it out to be. You talk about weather forecasting yet that isn't what we are talking about. We are ordering which objects to move in sequence for each to get to a known location most efficiently. With weather we don't get the ability to dictate what it does. It is known where the vehicle occupant wants to go. Even if the driver wants to change course that action is "known" and would need to be communicated either to a computer/network or in the case of a human driver through a series of actions to the vehicle itself. And still weather "prediction" has gotten quite good in the last few years and that's with many MANY more complexities than what we are talking about.


I like how you dismiss other points of view as people deluding themselves. A networked traffic system is one that would have to grow up. It wouldn't be something that you switch on and everything "works" And it most definitely wouldn't be "perfect" but then I don't think we are arguing that if it isn't perfect then it isn't worth it. It just adds to your game of dismissal.

We move insanely large numbers of packets of information across medium with near perfect results every single day. Thinking about traffic in the same manner isn't delusional or ridiculous at all. It's exactly how we should be thinking about transportation.

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 18th, 2015, 3:56 pm

Put it this way David. We are already doing this with rudimentary communication skills such as MAYBE turn signals and sometimes eye contact. To think that networked systems can't handle that type of communication to at first aid in our ability to drive and eventually to prioritize our driving might be a little bit arrogant.

Anondson
IDS Center
Posts: 4646
Joined: July 21st, 2013, 8:57 pm
Location: Where West Minneapolis Once Was

Re: Future cars / Driverless cars

Postby Anondson » December 18th, 2015, 4:40 pm

Weather patterns are much less predictable than cars traveling down lanes to a known location.
I'd wager weather is much more predictable than streets with pedestrian activity.

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 18th, 2015, 8:08 pm

Y'all missed my point. Ignore weather prediction as that's just an application. The problem is in keeping the hardware running and recovering from failure. Yes, we can make the network robust but information will be lost.

As I said, such communication will be useful, but I think people are greatly overestimating the impact it will have. It's not simply a matter of getting one car somewhere most efficiently. It's a problem of getting all cars everywhere most efficiently. It cannot be done in general. Heuristics will attempt to do it and will often work well but such things can also produce pathologically bad results.

It is *not* simple math. It's complex real-time big data analysis. This is the stuff of NSA and I actually doubt even they do it as quickly as would be needed to make a major impact on traffic flow.

Sent from my Z958 using Tapatalk

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 18th, 2015, 8:20 pm

The internet is an interesting analogy but I'm not sure it translates. Packets have way less latency than a vehicle. We do see packet congestion every day but the stuff moves so fast it appears to us as a reduction in FPS or some other inconsequential thing. It seems to me the impact on vehicle traffic would be much worse.

Even so, look at how we are still trying to optimize packet floor. We're in our 5th decade of that. It's been years since buffer bloat was identified as a problem and as far as I know we still don't have a widely deployed solution.

Look, I think distributed traffic management is worth trying and I'm sorry I was probably too dismissive. It comes from many years of hearing PRT bullshit from people who couldn't even address simple questions about how the system would work. I realize it was unfair of me to put that all on you, moda253.

I still think this stuff is horribly oversold, though. :)

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 18th, 2015, 8:36 pm

Continuing the internet analogy, we could choose to optimize for latency or throughput. Most drivers would prefer the former but the overall system is most efficient with the latter.

I admit I'm no network expert so correct me if I'm wrong, but my understanding is that traffic shaping and data prioritization is still a fairly big unsolved problem.

The internet has the advantage that once a packet is sent, its destination doesn't change. What about occupants who decide to change where they're going mid-trip? I think the internet probably has much more flexible routing and more path options than is available in a physical transportation system. It also had buffering resources that may not be available in the physical world. And of course the internet can and does drop packets. Frequently. Not an option on the road.

It may indeed be useful to think about this problem as a computer network but I'm not sure the same algorithms will apply and/or work all that well.

Sent from my Z958 using Tapatalk

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 21st, 2015, 9:25 am

I'm not saying it is incredibly simple but it is going to get increasingly easier over the next 5-10 years. And no it won't be perfect. It will be problematic, but from there on out it will become better than human drivers very quickly.

Take away the networked vehicle idea for a second and think about what they are doing with vehicles that operate as a single object. They already are pretty good at recognizing obstacles and navigating traffic. that is happening right now.... today. And it is still in its infancy stages. Now take that ability and apply sharing of that information to vehicles. At first while most vehicles are not outfitted with that type of tech it won't be very beneficial. BUt down the road when near every car on the road is sharing the information about an obstacle on the road you can have lane prioritization happening well before it becomes a problem. The transfer of the information miles previously is very possible.

The argument that it is too difficult right now to me is completely baffling. That's not even relevant to a conversation like this. It doesn't matter because in a few short years this type of computing is going to be child's play. And we are already on the verge of it. It's happening and it is already being worked on.

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 21st, 2015, 9:46 am

A bit cheesy and obviously produced to convey their agenda but this is a good watch...

https://www.youtube.com/watch?v=I2JY3rWzrqI

talindsay
Wells Fargo Center
Posts: 1533
Joined: September 29th, 2012, 10:41 am

Re: Future cars / Driverless cars

Postby talindsay » December 21st, 2015, 10:24 am

If you want to follow the internet analogy, please read up on packet collisions, the three-way handshake, and other ways of addressing packets that never arrive. The mid-level protocols (TCP/IP) are written in a way that you're generally blissfully unaware of the huge amount of data loss that happens in transmission. When every "packet" is a person it's a little harder to address those network problems: you can't just resend the packet.

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 21st, 2015, 11:31 am

If you want to follow the internet analogy, please read up on packet collisions, the three-way handshake, and other ways of addressing packets that never arrive. The mid-level protocols (TCP/IP) are written in a way that you're generally blissfully unaware of the huge amount of data loss that happens in transmission. When every "packet" is a person it's a little harder to address those network problems: you can't just resend the packet.

After working in tech for over 20 years I understand that. I would think that people would understand that analogies don't always have to carry down to the extremes of the concepts behind them. If you don't think that computer assisted and potentially computer driven vehicles isn't going to happen I don't know what to tell ya... It is going to happen. It has been happening for a long time. Since the advent of networked traffic signals. It is a sizable part of auto manufacturers focus right now.

I think we are getting lost on the idea that smart cars are going to somehow remove any and all problems related to transportation which I don't think anyone is arguing. Hey living is hazardous to your health. The question isn't can a computer drive your car better than you? Because driving better is subjective.

I know further back in this thread there was talk about cars driving themselves off to some storage location. And that's really interesting to think about happening someday. And that someday likely will get here faster than we think about it when these other systems that are emerging right now mature.

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 21st, 2015, 11:59 am

The argument that it is too difficult right now to me is completely baffling. That's not even relevant to a conversation like this. It doesn't matter because in a few short years this type of computing is going to be child's play. And we are already on the verge of it. It's happening and it is already being worked on.
Child's play? I don't think so. Again, you're talking about (probable) NP-complete problems. Those are never "child's play."

If such a system, once fully deployed, achieved a 20% improvement in congestion that would be phenomenal to me. Now I'm just pulling that number out of thin air but I just don't see this technology ever reducing congestion by even 50%. I hope I'm proven wrong and it would be silly to say it will never happen. But the challenges are absolutely enormous.

And again, that's after full deployment, meaning everyone has a self-driving car and ONLY operates in self-driving mode. I don't think that will happen within our lifetimes.

I think about self-driving cars being the congestion savior and the picture that comes to mind is the opening of Futurama with the traffic jam of flying cars.
I would think that people would understand that analogies don't always have to carry down to the extremes of the concepts behind them.
We're not talking about extremes. You're referencing absolutely fundamental properties of the internet when arguing that routing cars is just as "simple" as routing packets. If the internet ever lost the ability to drop packets it would collapse in a microsecond. I mean:
We move insanely large numbers of packets of information across medium with near perfect results every single day.
That's just not true if the analogy is made to moving cars. It is not even "near perfect" because of how much data gets dropped. When moving people you have to be absolutely perfect in that regard 100% of the time.

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 21st, 2015, 4:23 pm

The argument that it is too difficult right now to me is completely baffling. That's not even relevant to a conversation like this. It doesn't matter because in a few short years this type of computing is going to be child's play. And we are already on the verge of it. It's happening and it is already being worked on.
Child's play? I don't think so. Again, you're talking about (probable) NP-complete problems. Those are never "child's play."

If such a system, once fully deployed, achieved a 20% improvement in congestion that would be phenomenal to me. Now I'm just pulling that number out of thin air but I just don't see this technology ever reducing congestion by even 50%. I hope I'm proven wrong and it would be silly to say it will never happen. But the challenges are absolutely enormous.

And again, that's after full deployment, meaning everyone has a self-driving car and ONLY operates in self-driving mode. I don't think that will happen within our lifetimes.

I think about self-driving cars being the congestion savior and the picture that comes to mind is the opening of Futurama with the traffic jam of flying cars.
I would think that people would understand that analogies don't always have to carry down to the extremes of the concepts behind them.
We're not talking about extremes. You're referencing absolutely fundamental properties of the internet when arguing that routing cars is just as "simple" as routing packets. If the internet ever lost the ability to drop packets it would collapse in a microsecond. I mean:
We move insanely large numbers of packets of information across medium with near perfect results every single day.
That's just not true if the analogy is made to moving cars. It is not even "near perfect" because of how much data gets dropped. When moving people you have to be absolutely perfect in that regard 100% of the time.
Oh when will I learn David Greene He knows absolutely everything there is to know about everything... Just because.

I'll check back with you in 5 years time and see where we are. You are wrong on this David. Dead wrong.

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 21st, 2015, 4:31 pm

Whatever, man. This isn't personal.

Maybe I'm wrong. I'm comfortable with that.

Sent from my Z958 using Tapatalk

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 21st, 2015, 4:37 pm

If you don't think that computer assisted and potentially computer driven vehicles isn't going to happen I don't know what to tell ya... It is going to happen.
Yes, it's going to happen. No one here disagrees with that. The current discussion concerns whether automatic vehicles will have a significant impact on congestion. I'm quite skeptical about that. I though that you believe they will but the quote above makes me question whether we're even talking about the same thing.

Sent from my Z958 using Tapatalk

moda253
Nicollet Mall
Posts: 142
Joined: June 27th, 2012, 3:28 pm
Contact:

Re: Future cars / Driverless cars

Postby moda253 » December 21st, 2015, 4:55 pm

If you don't think that computer assisted and potentially computer driven vehicles isn't going to happen I don't know what to tell ya... It is going to happen.
Yes, it's going to happen. No one here disagrees with that. The current discussion concerns whether automatic vehicles will have a significant impact on congestion. I'm quite skeptical about that. I though that you believe they will but the quote above makes me question whether we're even talking about the same thing.

Sent from my Z958 using Tapatalk

If you look back you will see that it my stance is that it isn't going to happen overnight. We didn't start with the internet. We started with pony express and then the telegraph, ma bell, aloha net..... of course skipping a lot of things in between but you get the point.

Driver assistance is here. now. it exists. The standard for vehicle communication exists in 802.11p and DSRC for V2V and V2I is in place already. Much of that has been in place for several years now. Market turnover will be the biggest piece of the puzzle in getting it up and running. But then you have cheap aftermarket systems that can be sold to aid in that as well for people to buy into the system.

So we have assisted cars already here and they have been here for a few years now. Automated Cars are a thing which ARE going to be on the road probably by 2020. That is happening. V2V (vehicle to vehicle) and V2I (vehicle to infrastructure) already has adopted a standard and is operating in it's own communications band so that is up and running.

The next phase of that is going to be perfecting that system to a point where it is reasonable to think that cars can for the most part drive themselves unassisted. Right now that is a big leap for people to believe in.

Then again years ago if you left the coffee pot on you'd go racing back home to turn it off. Now people rely on auto-off to do it for them. And because undoubtedly someone is going to point out the mundane and ridiculously obvious differences between a coffee pot and a car. no I am NOT SAYING THAT CARS ARE THE SAME THING AS COFFEE POTS!

User avatar
VacantLuxuries
Foshay Tower
Posts: 973
Joined: February 20th, 2015, 12:38 pm

Re: Future cars / Driverless cars

Postby VacantLuxuries » December 21st, 2015, 5:03 pm

Self driving cars will have zero impact on congestion until they're the only kind of car. The human element has always been the issue with cars - engineers design freeways like a network of pipes, but humans don't behave like water. They change their minds, they operate outside the law, they make rash decisions based on wanting to get to an exit ten seconds faster.

So I do think self driving cars do have the potential to fix congestion. But I have a feeling there's a sizable segment of the public that will fight full automation tooth and nail, whether they're an automaker who doesn't want to see their product lose it's luster, someone whose income depends on driving remaining a skill, or just someone who doesn't want to give up on the notion of the car as a sexy power symbol they can drive above the speed limit in. Once you get past all that, maybe self driving cars will solve congestion. But expect the fight to be as impassioned as gun control when it finally arrives. Possibly moreso, since car culture is arguably more mainstream than gun culture.

User avatar
Nick
Capella Tower
Posts: 2719
Joined: May 30th, 2012, 9:33 pm
Location: Downtown, Minneapolis

Re: Future cars / Driverless cars

Postby Nick » December 21st, 2015, 5:08 pm

I am NOT SAYING THAT CARS ARE THE SAME THING AS COFFEE POTS!
https://streets.mn/2015/01/08/cars-as-keurigs/
Nick Magrino
[email protected]

David Greene
IDS Center
Posts: 4617
Joined: December 4th, 2012, 11:41 am

Re: Future cars / Driverless cars

Postby David Greene » December 21st, 2015, 8:34 pm

The question of whether driverless cars will "fix" congestion is actually a very important one right now. The same policy-makers who say driverless cars will fix our transportation problems are pretty much exactly the same people that predicted PRT would do the same, using the exact same reasoning (automation = perfect outcomes). In other words, driverless cars are the new wonder cure to justify cutting funding for transit.

I think we all agree that even if dirverless cars achieve some unprecedented breakthrough, it won't be for quite some time. But we need transportation fixes *now.* As others have said, such a breakthrough would still leave transit on top as far as efficiency, reduced climate impact, etc. But regardless, I do not believe driverless cars are a congestion solution.

In any case, we have to nip the driverless cars > transit narrative in the bud. Stop it now before it gains legitimacy and stalls legislative action for another decade. Because the climate can't wait for yet another technology that is "only a decade away."

I hope we can all at least agree on that.


Who is online

Users browsing this forum: Bing [Bot] and 57 guests