self-driving cars

 So you noticed these guys this is 2022. And these general motors guys and I don't know maybe I don't know who else but certainly Tesla and this guy George. These guys keep saying they're going to have a fully autonomous dry you know driving car that can drive anywhere in the country and then they keep liking 2016 I think they started saying that by 2017 they were going to have it. And then they changed it till like 2018 and 19 and 20 and 21 and 22. And now I think they're saying like by 24 or 25. So if you look at what they can do that they all seem to be able to kind of do the same thing around the same time even though they're all working independently but they all watch each other and and then they go oh we need to make ours do that. And Tesla has kind of been the the car to be. So it's kind of a strange competition right where they're all competing against each other but they're not really competing against the problem it doesn't seem like. Now Tesla I I guess is, and I've heard these words come out of out of elon's mouth he goes with a problem is harder than we thought. Well the problem of automation has been obvious now for I don't know the last 60 years at least so people are not so that they're I don't know what's wrong with us. A self-driving car if if you look at George's solution which is a forward-facing camera that attaches to your windshield and then hooks up to a bunch of servo motors in the car and that thing is getting to be pretty good it it can come close to rivaling Tesla's thing. His thing will fit in somewhere around 100 different models of cars and it's about $2,000. Now it still doesn't just drive the car all by itself. But it does pretty good. So they need to there's still some fundamental change in perspective that they need. Like make the car stay on you know flat open surfaces and it don't hit anything else so that you know if you just came up with some simple rule like that better that the thing had to do it had to you know so of course it has to go up small inclines and things like that in it but you didn't but we're just going to say that for this exercise you never go over curbs you do have to go over speed bumps. Avoiding stuff garbage and obstructions on the road 2x4s and stuff that can damage the car that's certainly a big nice to have but it has to completely like absolutely not hit other cars or not hit people. So even those kind of minimal rules seems you know very difficult. Just drive on flat surfaces and stay out of trouble.

One of the problems is that taking a live feed video and turning it into commands to drive the car is a pretty huge jump in processing. You need to take to some degree still life photos that are coming in at somewhere around 33 frames per second and you need to take those and then determine like where to drive the car. So that's that's a pretty you know a processor intensive thing. But just looking at a picture I'm at I think I could get the computer to look at a picture and tell me where I could possibly drive right the rule is if there's an open space stay on the right and don't hit stuff. Now of course the first problem is when you get the first exit you just always take that exit right cuz you're staying to the right. And and that's exactly what these guys find the thing won't stay on the road or the thing stays on the road all the time and won't change lanes ever right side it's it's sweet we keep wanting some sort of simplistic you know just make the car go from point a to point b but I mean that's not what humans do humans do tons of processing.