Hacker News Reader

Comment by adrr | Hacker News Reader

If cars could talk to each other you'll get better utilization of the road because you could cut follow distance to a few feet.

View on HN

Sounds good in theory. I’m not optimistic about all the car makers cooperating on an industry standard, though. Plus the failure scenario seems pretty catastrophic to overall throughput. I hope some bright thinker figures those problems out; it does seem like a great opportunity

It doesn't seem like you'd need an "industry standard" beyond the actual rules of the road. In the same way that normal human-operated cars with better visibility can change lanes on the highway more safely and efficiently, any improvements in sensor ability, reaction times, etc. in self-driving cars (and even automatic safety features for human-operated cars) should make traffic safer and more efficient.

what do the actual rules of the road say the braking point should be for an 8 person minivan with 5 passengers in dry weather? what about 8 passengers in the wet at night with ice on the roads?

old argument i know, but if a human gets that wrong and plows a pedestrian they wear the consequences, will an auto manufacturer do the same?

my guess is no, and the result of that is that even if we can get a majority of cars to be self driving they are probably not going to be pushing the limits of road capacity any time soon like GP suggests, more likely they will continue to drive super conservatively and fail fast like the videos in the linked article.

No, because physics is not affected.

Instant reaction time to what the car is doing in front of you. What physics are you talking about?

Say the car in front of you runs into a stationary object, or something travelling the other way. Instant reaction time wont help as there is no way to decelerate as fast as a colliding car.

Stationary object? Like a tree or a light pole?

More like a Tesla that randomly freaks out and emergency brakes to a dead stop in the middle of 70mph traffic for absolutely no reason at all.

As happens surprisingly frequently.

Wouldn't the car in front detect the stopped tesla, the lead car would hit the brakes and the cars behind it would hit their brakes at the exact same time?

I'm not convinced it would detect it significantly faster than a human.

We are surprisingly fast, and the slow part is that we have to perceive that the thing in the road is no longer moving past the landmarks in the road.

Assuming most cars have similar stopping distances. There would be no rear end collisions if we had instant reaction time to cars stopping in front us. Rear end collision is the most common collision and attributed to follow distances. Standard safe follow distance is 3 seconds. Standard reaction time is 1 second which means at 60mph you travel 88 ft. Most modern cars can stop from 60mph around 150ft.

https://www.qld.gov.au/transport/safety/road-safety/driving-...

Reaction time isn't really significant.

The variation in brake performance between vehicles will more than outweigh the effects of reaction time.

My old Citroën CX will easily outbrake a Tesla. If you are driving too close, even if the Tesla could brake at the same instant as I do, it will still hit me.

How long is it going to take to communicate between cars? How do you see that communication working? How will it respond to interference and packet loss?

https://www.qld.gov.au/transport/safety/road-safety/driving-...

Which just proves my point. Shaving 1% off the total stopping distance by making it react a little quicker isn't going to help.

You mean 1 second which at 60mph is 88 feet. Tesla stopping distance at 60mph is 150 ft for comparison.

Reply on news.ycombinator.com