Hacker News Reader

Comment by darth_avocado | Hacker News Reader

I was told by a very intelligent man demanding a trillion dollar salary that you only need vision cameras to have full self driving in all weather conditions. All of this is apparently unnecessary.

View on HN

I don't think either solution is going to be the eventual winner. I expect the eventual rollout of 5G+ to be the game changer.

Much like how a pilot captain boards a ship to steer it into port, traffic systems will be able to be a bit more 'hands on' when it comes to getting traffic through junctions safely, regardless of the weather. Hence, on the final journey through a city, the city traffic systems will be directing your car.

For the journey on highways between built up areas, variants of today's self driving systems will suffice.

Lots of people manage to drive in poor visibility without radar, lidar, etc. If that's safe and normal for people, the principle that self driving should work with just cameras isn't such a strange one.

I wouldn't call things like radar and lidar unnecessary but in principle a good AI vision system should be able to operate at the same level as a human eventually. Of course if you don't have such an AI just yet, you need a stop gap solution. But I wouldn't bet against AI getting there eventually. Probably not on Mr. Musk's accelerated and optimistic schedule though. But give it another five to ten years and things might look a bit differently.

What Waymo is doing now with much less than perfect AI is of course completely pragmatic and very impressive. I'm kind of eager to see them start operating self driving outside a few restricted zones in the US and for example in European cities. I live in Berlin, so probably we'll have to wait quite a bit for people to finally let go of their fax machines though apparently there are some trials with self driving buses about to kick off here now.

> Lots of people manage to drive in poor visibility without radar, lidar, etc. If that's safe and normal for people, the principle that self driving should work with just cameras isn't such a strange one.

No camera system comes close to the capabilities of human eyes, combined with general intelligence.

I am one of those who drive in bad weather from time to time. I'm 'good at it' - but I cannot honestly call myself safe. I've been in the ditch. I've spun a 360 and only didn't hit someone else in the process because it happened nobody was there.

i grew up where bad weather was common enough that we cosidered it not worth shutting down for bad weather so we risked driving in it - but it was always a risk and many do die from taking that risk.

Just vision is a very reductive way of describing how we drive. We use sound to hear if something else is coming and where from, or if we're losing traction on the road, or if the road surface has changed. We can judge acceleration and deceleration in all directions using our inner ears. We can feel if the car's performing differently to usual indicating an issue with the car or different conditions. In addition when we are using our vision, if it gets obscured (i.e. snow covers the windscreen) we know how to get it off with wipers, and most importantly we're very adaptable to new conditions in a way that computers aren't, if we experience something completely new to us, chances are we'll make a reasonable in the moment decision.

The vision-only approach surely seems to be falling behind the multi-sensor approach.

The vision-only approach isn't even able to trigger the windshield wipers correctly.

This annoys me so much when it rains. I dont understand how tesla is still selling any cars outside of arid climates

Same - living in a region where it gets below 0°C in the winter and also snows, you really see that those cars were designed for sunny california. It happened multiple times now that the trunk or door handles were frozen so hard, that I couldn't use the car in the morning. Also when opening the frunk while its partly covered in snow just made it slide into the frunk...

It's fine to dislike the guy, but stock options aren't the same thing as a salary.

they are when you can open a line of credit against them

No? Words mean things. Just use the right words.

Sure. But when the option is priced at strike price way below the market price, it can be pretty much considered as salary.

It's meaningfully different. Salary is a fixed, regularly recurring income stream. That isn't what these are. This is compensation and it has cash value, but not all compensation is salary.

He is not wrong, but we demand superhuman performance from our machines which in this case necessitates superhuman sensory abilities. Current evidence shows that having non-vision sensors is a faster way to create a reliable system. I would personally choose to ride in an autonomous vehicle with Lidars.

It seems quite likely that once self driving cars are well perfected, we will demand more than just human level driving which is currently horrendously dangerous. If lidar systems can exceed vision only, we are going to demand it as a baseline standard.

> He is not wrong, but we demand superhuman performance from our machines

I have a model 3 with v3 FSD hardware. FSD is an objectively terrible driver compared to the average human.

And if the US government operated with 10% of the agency or spine it ought to operate with, the entire feature would be banned and tesla fined for costing so many lives already. And the cyber truck wouldn't be coasting the roads with no safety sense whatsoever.

He is very obviously wrong since Waymo cars drive millions of trips with 0 drivers while every single robotaxi still has a safety driver in it at all times.

"He is not wrong, but anyway, here's why he's wrong."

He's wrong. Cameras are not enough... but they're certainly cheap enough.

Reply on news.ycombinator.com