Yes, I meant to imply that you’ll have a semi-AV in your lifetime. Apparently you already do, but it sucks. The current gen Tesla models can operate okay in inclement weather, and they can do highways and towns and dirt roads too. So I think they’d be almost useful for you. Maybe not quite, but within a decade.
They fail specifically in large cities with complicated lanes and signage. They also currently fail at taking unprotected lefts on busy streets. They have pretty specific points of failure, and they really are gradually improving.
I also suspect Semi-AVs will be ubiquitous among new cars, pretty soon, so we’ll all probably own one eventually. The key components-- cameras and computer chips-- are cheap to manufacture. And they will prevent accidents.
(That isn’t to say that all the cars will be AV. I don’t think we will outlaw old cars for many decades.)
I also think more serious AVs are coming, but I can’t guess when, or where, or what they’ll be used for. They might suddenly be everywhere, or they might be geo-fenced, special purpose, etc.
I mean, it’s nice when it works, but it fails way too often to be something you can rely on.
I certainly made use of it when making the 1,000 mile trek to/from my in-laws. It was noticeably less fatiguing than driving to the (slightly closer) cattle ranch in my old car, which was my previous long road trip.
All this talk about self driving cars and the use if interstates has raised a question and I am sure that there is a simple answer that I just don’t see. Let’s say we have 3 lanes highways (3 in each direction) and my car is on the highway. What is the protocol for lane selection? Assuming that there is no speeding, so every car is going the same speed, what are the criteria for lane selection? Is it miles to the exit?
What’s the confusion here? I’m sure this can be worked out easily with an algorithm.
The efficiency of an AV doesn’t come from the automation from a singular source, it’s from the collective. At least, I believe auto makers will realize that and be willing to collaborate with each other.
A single car responding to its environment is okay, but if every car can communicate and access the data (relevant data) of every other car, then that’s where efficiency is maximized.
In this scenario, every car will know which cars are nearing their exits, and will start to swap lanes simultaneously without much change in velocity.
I have EyeSight on my Subaru and it regularly craps out at night or in the rain, in conditions where I can still see pretty well. There is no way in hell self-driving cars are withing five years of being generally on the road.
There are more fenderbenders on surface-level streets, of course, but the interstates are where people die. I could see this one both ways through different lenses.
IIRC, self-driving capabilities are present in long-haul trucking, and can generally be used when on the interstate, but not on roads with traffic signals and stop signs.
So for private passenger type vehicles, I see a similar sort of situation developing.
Driving on “surface level streets” is much more challenging. And it’s also much harder to upgrade them (all!) to talk to autonomous vehicles. I expect caravans of mostly-self-driving trucks to appear on the interstates years before there’s any economically significant autonomous driving on surface streets.
You people must not live in deer/elk/moose country. I think keeping the car moving forward along a normal path without hitting other traffic is probably easier on the interstate than city roads. I think the type of obstacles to identify and avoid on the interstate are much more complex. Trying to identify a deer standing on the edge of the road standing still that will jump out at the last moment is not an easy task. Nor is detecting a trailer hitch in the middle of the road while traveling at 70 MPH. The slower speeds of surface roads provide longer reaction times which I think will result in city driving being “solved” before interstate driving.
That being said, a big truck can turn a deer into paste without leaving any damage on the truck so I agree that self driving interstate trucks are more likely to appear standard before other sefl driving vehicles.
“reaction time” is actually one of the areas where computers totally outclass people. The question is more about how hard it is to get the answer right. And how bad it is if you mess up. There are a LOT more pedestrians on surface roads.
I think you are underestimating how long it takes an AI process to discern the difference between a shadow on the road and a trailer hitch. Regardless, it is much more likely that the AI comes to the right decision when moving at 35 MPH versus 70 MPH. Pedestrians are also much easier for AI to recognize because they are moving which triggers the AI to “pay attention” and they tend to not suddenly jump into the road because people are semi-rational. Once again, the lower speed on surface roads translates to a higher chance that the AI makes the right decision. You also have never seen a moose go through a car windshield. Personally, I find night driving on country roads much more difficult and taxing than city driving because you have to be so focused on what is at the side of the road. On the other hand, I have almost no experience driving in very large cities so maybe driving in a place like New York City or Los Angeles is very different than cities I have driven in.
navigating the high way is way easier than a costco parking lot