Hacker News Reader

Comment by JoachimS | Hacker News Reader

I'm probably dim. But I fail to understand why self driving cars will make the congestions in morning and evening commutes be radically reduced.

If the wall described in the article is to be pushed out, the speed must increase from today.

If we can go to work and go home at more varying times over a day then yes congestion should decrease. But why would AVs cause that effect? As long as everybody is expected to be at the office at 8am and leave at about 5pm, we need to move the same amount of people back and forth at the same time. Ride sharing (hey buses and metros are efficient for that!) may make the number of vehicles/person go down.

View on HN

The amount of vehicles a road can handle before becoming congested is dependent on the safe distance needed between cars at a particular speed, the reaction time of drivers, and the consistency and coordination of driving styles.

You won't see any improvements from AVs until human drivers are banned, after which the AVs can all drive at 100 mph and inches apart from each other. Traffic lights would also no longer be necessary.

Is human reaction time is the only reason for enforcing safe driving distance?

Even if AI increases to great levels, inertia still would exist. A large truck cannot stop in inches when traveling at 100 mph.

Human reaction time would be a reason for additional distance beyond what is needed to stop, but agreed, I don't see the distance to be reduced that much. Reducing the distance to virtually nil is going to cause a crash at the first slope as well.

If the reliability is high enough then the low rate of accidents is worth the risk. Other modes of transport are able to operate at small margins due to high reliability requirements.

It only has to stop as quickly as the truck in front.

An interesting thing about objects is that they can move in three dimensions.

All you need is an object of any kind, including a downed tree limb or falling rock, entering the stream of vehicles at an angle, to cause cataclysmic disaster in this scenario.

But the truck in front busted a tire... Now what? If the distance between vehicles is 1cm,then literally thousands of vehicles will crash.

The AVs could offer some advantages (ie cheap car sharing via resource pooling, reduced parking space requirements in urban centers) but safety or speed aren't part of the deal. AI isn't superhuman and AVs aren't able to brake the rules of physics.

what if the vehicle in front is a small car?

You forgot the most important factor: the exits. If the exits are congested because capacity is lower near the destination, there's nothing you can do in the road.

And most jams are caused exactly by this.

> AVs can all drive at 100 mph and inches apart from each other

Hard to see how this could ever go wrong isn't it

You would also have to ban pedestrians and cyclists.

And you shouldn't care about noise anymore. When there's a constant stream of self-driving cars at 100 mph on the streets every single place in the city will sound like living next to a major highway.

And no, electric engines won't solve this problem, at speeds above 30 mph the engine is no longer the major source of noise, it's the tires and the aerodynamic noise.

It's not viable in cities, but large parts of commutes and congestion are on freeways where they are banned anyway. We don't even want cars driving 100 mph in cities because of noise and pollution (for gasoline cars).

And if we went fully AV, we could still do away with traffic lights for cars in cities, as long as pedestrians and cyclists have their own safe passages and/or AVs are smart enough to deal with them just crossing at will.

after which the AVs can all drive at 100 mph and inches apart from each other

Except... it's likely that people won't own AVs of their own, so we'll be captive passengers inside a corporation's AV. That's a prime advertising opportunity. Why have an AV that gets people from A to B as quickly as possible when you can go at 55mph without needing to change the law and get double the time with the passenger?

Good point, though I suppose it would be a battle between corps that need productive time from their employees rather than time watching ads in a car. I guess if the time is stolen from personal time rather than work time then all good.

Traffic lights would also no longer be necessary

How will pedestrians know when to cross the road? Will bicycles also become illegal?

>You won't see any improvements from AVs until human drivers are banned

Automatically re-routing traffic when the roads start being congested already works before that and it'll help a lot.

Soon mankind will realize that it is better to spend one dollar and remove all ads rather than imposing major annoyance in the hope of gaining 5 cents.

They won't, it's something AV proponents love to claim though.

The claim is that cars will be able to drive closer at higher speeds or something. But, 2 tons of steel is still a deadly weapon no matter who's driving it. A car can still suffer a(n unpredictable) blowout or mechanics failure, malicious people can still screw with traffic, roads and infrastructure will still be poorly funded, and even wildlife (deer specifically) still exist.

Any of these things will cause huge problems to cars travelling too close together. That's even ignoring the fact that traffic going inbound to dense urban centers is limited by the infrastructure of the city and layout of those roads, parking garages, and businesses- not the speed at which people can travel closely on the road.

By the time we get level 5 autonomous vehicles, we'll have robots performing most minimum wage jobs (picking crops, making coffee and hamburgers, retail) and culture will probably have shifted to a remote-work-friendly mindset. The fact that there just won't be that many jobs for people to work will reduce traffic more than AVs will "because they can go faster".

Numberphile had a recent video on car crashes here: https://www.youtube.com/watch?v=i3D7XYQExt0

In short: Since kinetic energy goes as (velocity)^2, doubling the speed will quadruple the stopping distance (MANY other factors are significant, this is a rough estimate). 3X of the speed is ~9X the stopping distance, etc.

NHTSB guidelines for 'proper' following distance are nearly never seen in LA and NYC, it's typically much shorter. If anything, L5 AVs will probably have longer gaps between AVs than what is currently on the road. Likely, the major improvement of L5 AVs will be in safety, not congestion. It turns out (I think), you can't technology your way out of a political problem. We need higher density housing, not magic carpets (effectively).

Reply on news.ycombinator.com