4 ms·

I don't get what's so difficult about self driving. The car just needs to show a path it's going to follow during the next ten seconds and stick with it. Only w

I don't get what's so difficult about self driving. The car just needs to show a path it's going to follow during the next ten seconds and stick with it. Only when the path appears wrong, the driver takes control. This let's the driver know how the car sees the road. Unless the self driving saga has a different goal: transportation as a service. This is where the money are and this is where full auto driving is a necessity.

camjohnson26 6y ago
What’s so difficult about intelligent language models, you just pick the next few words the model will say.
- karmicthreat 6y ago
  Yea, really. What's so difficult about X usually means you have not thought about the problem enough. What's so difficult about making an app? You mean other than having to have some understanding of 20 years of tech its built on top of?
sfjailbird 6y ago
> The car just needs to show a path it's going to follow during the next ten seconds It could even project this onto the wind screen. Some new cars do a neat trick with perspective adjustment so the projection appears 'flat' even on a tilted wind screen.
pas 6y ago
Their world model is shit apparently. It should be a hierarchical Bayesian model, constantly updated, and checked for likehood distance from priors, and when it gets too big, alert the driver and disengage. Cars don't randomly appear and disappear in the middle of the open road. If their model cannot interpret these events as "a big fuckin' problem, I'm out of my operational envelope, I need to stop", then that model is indeed dangerous. Sure, this is easier said than done.
- frongpik 6y ago
  The ML model may be shit and that's ok so long as it shows the path it's going to follow and sticks with it.
  - pas 6y ago
    Sure, if it's simply a follow the car in front of me model. (Which is lightyears from FSD.) Also then it needs to be super-super finicky about what it interprets as a car/road/etc, and start disengaging the instant it detects something is not right. But obviously Tesla/Musk decided to just tone down the carefulness.

camjohnson26 6y ago

What’s so difficult about intelligent language models, you just pick the next few words the model will say.

karmicthreat 6y ago
Yea, really. What's so difficult about X usually means you have not thought about the problem enough. What's so difficult about making an app? You mean other than having to have some understanding of 20 years of tech its built on top of?

sfjailbird 6y ago

> The car just needs to show a path it's going to follow during the next ten seconds It could even project this onto the wind screen. Some new cars do a neat trick with perspective adjustment so the projection appears 'flat' even on a tilted wind screen.

pas 6y ago

Their world model is shit apparently. It should be a hierarchical Bayesian model, constantly updated, and checked for likehood distance from priors, and when it gets too big, alert the driver and disengage. Cars don't randomly appear and disappear in the middle of the open road. If their model cannot interpret these events as "a big fuckin' problem, I'm out of my operational envelope, I need to stop", then that model is indeed dangerous. Sure, this is easier said than done.

frongpik 6y ago
The ML model may be shit and that's ok so long as it shows the path it's going to follow and sticks with it.
- pas 6y ago
  Sure, if it's simply a follow the car in front of me model. (Which is lightyears from FSD.) Also then it needs to be super-super finicky about what it interprets as a car/road/etc, and start disengaging the instant it detects something is not right. But obviously Tesla/Musk decided to just tone down the carefulness.