Fix traffic: No more monkeys driving cars

[PhilDawson8270]

View: https://www.youtube.com/watch?v=iHzzSao6ypE


A very interesting video from a YouTube channel I subscribe to. Explains the causes of the mysterious no cause traffic jam, and looks at the issue with traffic jams.

Doesn't cover too much about cycling, and although the end is well thought out and seemingly a good idea. It appears that the author hasn't considered cycling at all.

So, I thought it would be interesting for their to be a discussion on where bicycles would fit in to the scenario at the end of the video? Would computers be more or less forgiving to us cyclists compared to humans?

 

[DCLane]

So basically, don't let people drive cars, let computers do it.

 

[phil_hg_uk]

The silly thing is my computer is quite happy sitting on its desk. Why would it need a car?

How do you know its happy. maybe it is really wanting to elope with the TV !!!!

 

[NorthernDave]

The silly thing is my computer is quite happy sitting on its desk. Why would it need a car?

Surely once one of them gets a car they'll all want one - and where will that end?

 

[mjr]

So, I thought it would be interesting for their to be a discussion on where bicycles would fit in to the scenario at the end of the video? Would computers be more or less forgiving to us cyclists compared to humans?

Cyclists are humans so I don't understand that second question!

The main trouble with that video is that there will always be someone (whether manufacturer or individual) who subtly adjusts their car-computer's attitude to be a little bit more Must-Get-In-Front than the others and exploit whatever flaws there are in their programming. That's the bit that self-driving cars seem to fail to address: they're just a faster and more reliable version of human-controlled motoring and the same systemic beggar-thy-neighbour flaws will remain in the system. It's essentially not sufficiently in your interests to serve the greater good. For a while, sure, computer control will enable cramming more cars into less space, but when it fails, it will be EPIC.

A serious flaw with that video's advice to human drivers to drive "in the middle" is that if you are being tailgated, you shouldn't speed up to tailgate the vehicle in front - and in the UK and many of our neighbours, if you merely seek to hold a two-second gap to the vehicle in front (in the interests of safety) and avoid extreme braking or acceleration (to improve economy of your own vehicle), then you'll probably be tailgated a lot and occasionally overtaken or crossed dangerously enough to require hard braking. And why not? Whoever causes you to brake and start a traffic compression doesn't get punished by it, do they? And it's not like there are many traffic police or insurers taking dashcam reports seriously now, is it?

 

[swee'pea99]

That is interesting, and certainly tallies with thoughts I've mused over, typically while stuck in traffic, that ultimately (by which I mean within my kids' lifetime, or at least within theirs') all traffic will consist of driverless cars coordinated centrally by a mega-computer. Drivers will not be allowed. Instead, the mega-computer will organise everything, nationwide, so's to optimise journeys for everyone.

You get in your car, say you want to go to 13 Acacia Avenue Luton, and away you go. The computer calculates how best to incorporate your journey into the overall 'journey matrix' currently in progress, optimising your journey so as to maximise your progress while minimising your impact on others', updating this journey plan in real time as other cars join/leave routes that intersect with yours, slowing you down and speeding you up as necessary to minimise or even eradicate stops, and getting everyone to their destination as quickly as (overall) possible, given the ultimate limiting parameter: the total road network.

Of course in reality, the system won't optimise travel for everyone. It'll prioritise, depending on ability/willingness to pay. Well, someone has to pay for the mega-computer...

 

[PhilDawson8270]

Cyclists are humans so I don't understand that second question!

I mean human drivers vs computer drivers with cyclists. The overly dramatic animation at the end. What would it be like trying to cycle through that?

 

[mjr]

Of course in reality, the system won't optimise travel for everyone. It'll prioritise, depending on ability/willingness to pay. Well, someone has to pay for the mega-computer...

And so, some who can't pay for what they feel they deserve will crack the system and instead of it being servers crashing and packet retries, it'll be motor vehicles crashing and unwary occupants dying. The day they force all automated-driving cars (they won't be "self"-driving any more) to link to one scheduling system is the first day that anyone with a self-preservation instinct should probably cycle to work/shops/whatever instead!

 

[mjr]

I mean human drivers vs computer drivers with cyclists. The overly dramatic animation at the end. What would it be like trying to cycle through that?

Bloody impossible to do safely, just like crossing the road. You'd be asking people to cycle/walk into live traffic, trust that everything would detect you correctly and obey the advertised rules. One acceptable solution is to limit the vehicle density so that there are sufficient gaps but that would undermine the "benefit" of nose-tail pollution-creation.

 

[DaveReading]

A very interesting video from a YouTube channel I subscribe to. Explains the causes of the mysterious no cause traffic jam, and looks at the issue with traffic jams.

Should be required viewing for all those moronic radio traffic reporters who persist in telling us that "everyone is slowing down to look at an accident".

 

[Drago]

The silly thing is my computer is quite happy sitting on its desk. Why would it need a car?

Think Man, think. It didn't walk to your desk, did it?

 

[marinyork]

I mean human drivers vs computer drivers with cyclists. The overly dramatic animation at the end. What would it be like trying to cycle through that?

Hard to say, obviously very early days with regards to machine learning. There's the famous google car that got confused by a trackstanding cyclist.

Jockeying causes a lot of problems in queues in various situations such as in supermarkets.