Meanwhile, the only actual L3 capable car is a Mercedes. Plenty of people will argue that AP is actually superior, but I don't think there's any way to say that for sure, given the tight constraints Mercedes has put on their driver assistance in order to lower the liability enough to make L3 not a bankruptcy event. For certain, however, they are more confident in their technology than Tesla is, otherwise Tesla would release an L3 capable car too.
To be fair, though, Tesla has no sensors other than cameras, and I believe the Mercedes has a half dozen or so, several radars and even a lidar.
> To be fair, though, Tesla has no sensors other than cameras, and I believe the Mercedes has a half dozen or so, several radars and even a lidar.
This doesn't mitigate Tesla's gross ethical violations in letting this thing loose, it makes them worse. Tesla knew that cameras alone would be harder to make work than cameras plus radar plus lidar, and they shouldn't be lauded for attempting FSD with just cameras. It's an arbitrary constraint that they imposed on themselves that is putting people's lives in danger.
Remember that big announcement musk did where he stated, that going forward, Teslas would not use any additional sensors to aid the video---to the point of laughing at every other company that was still using them. Yeah.
This is limited to certain pre mapped roads, under 40 mph and requires a car in front to follow. This in no way compares to Tesla's Autopilot system.
Edit (because HN throttling replies): > Mercedes is being responsible about their rollout
You can always succeed if you define success. If rolling out something that doesn't do much of anything is success, success? Oh well. Less than 20 people have died related to Autopilot in ~5 billion miles of travel. Zero deaths is unattainable and unrealistic, so "responsible" and "safe" is based on whatever narrative is being pushed. 43k people died in US car accidents last year, roughly half a million deaths in the entire time Tesla has offered some sort of driver assistance system called Autopilot.
How many deaths would be be okay with Mercedes' system where is would still be "responsible"? Because it won't be zero. Even assuming 100% penetration of Automatic Emergency Braking systems, it's only predicted to reduce fatalities by 13.2% (China specific study), and injuries by a similar number.
TLDR "Safe" is not zero deaths. It is a level of death we are comfortable with to continue to use vehicles to travel at scale.
Mercedes is being responsible about their rollout, so people like you jump to the conclusion that their tech is worse. For my part, I see Mercedes taking liability for accidents that happen and think that they must have a lot of confidence in their tech to roll it out at all under those terms.
I would not be surprised to see Mercedes beat Tesla to safe fully automated driving because they took it slow and steady instead of trying to rush it out to legacy vehicles that don't have the right sensor arrays.
Edit: It's weird to reply to a comment by editing your own. It feels like you're trying to preempt what your interlocutor is saying, rather than reply.
It is in no way comparable to Tesla's Autopilot system. Mercedes has an actual feature while Tesla recklessly gambles with other peoples' lives to juice their stock price.
If Tesla believes that their feature is safe, then let them take legal liability for it.
The law does not require they take liability, so they shouldn’t. If you don’t like the law, change the law. If you don’t want to use a driver assist feature, don’t. But the data shows Autopilot is robust and regulators allow its continued sale and operation.
>If you don’t want to use a driver assist feature, don’t. But the data shows Autopilot is robust and regulators allow its continued sale and operation.
uhh,, this is a thread where the federal gov't just forced Tesla to patch the feature, and so far we don't even know what the patch does...
Calling it L3 capable is like saying Tesla is Full Self Driving.
Neither is true, both are marketing.
> but I don't think there's any way to say that for sure
I mean yes there is. If you want to have an object test where you drop a car anywhere in the world and see if it can get somewhere else, then clearly one is more useful then the other.
In basketball they say 'the best ability is availability'. In terms of that AP is in a totally different dimension. AP has been driven for 100s of millions of miles by know, it must be a crazy high number by now. Mercedes L3 system has barley driven at all, its available in very few cars.
The only way you can reasonably compare the Mercedes L3 system is if you limit the comparison to the extremely limited cases where the Mercedes L3 system is available. If you compare them there, I would think they aren't that different.
> otherwise Tesla would release an L3 capable car too
No, because making something L3 in a extremely limited selection of places is simply not something Tesla is interested in doing. Doing so would be a lot of work that they simply don't consider worth doing when they are trying to solve the larger problem.
If Tesla's work is actually going somewhere, then they don't need to change the engineering to get L3 in specific places. They can have a list that gets added to over time.
Actually it would be a lot of work doing that. Even if not just work on the software itself. Something like that would cause a lot of work in literally all over the company.
Also, they don't really have an inattentive to do that, so why should they.
Why? The design is locked in, the manufacturing doesn't change, you don't need the designers to do much... I would expect you need a handful of people, half of which are lawyers.
> Also, they don't really have an inattentive to do that, so why should they.
I'm pretty sure almost all Tesla owners would like the ability to stop looking at the road some of the time. That amount of customer satisfaction is not a motivation for the company?
That Tesla was allowed to test in production without any legal liability with real human lives mostly because of Musk's personal influence is a disgrace
> That Tesla was allowed to test in production without any legal liability with real human lives mostly because of Musk's personal influence is a disgrace
I am unsure what"allowed" means in that context. They just did it. The lawsuits are coming, surely?
Sometimes it is better to ask for forgiveness than for permission. This was an audacious example. Time will tell if there is anything that will stop them.
It's available in bumper to bumper rush hour traffic on freeways, and they take liability for any accidents that happen while it's on. That's exactly the kind of system and guarantee that could really have a positive impact on someone's commute.
Don't LIDAR/radar sensors (I'm not familiar with exactly what the options are) have benefits that vision doesn't have, like working in poor lighting/visibility? Why would Tesla move away from these sensors?
Lidar has advantages over cameras but it also has some downsides. Sunlight, rain, and snow can interfere with the sensors, as can other nearby lidar devices (though de-noising algorithms are always improving). There are also issues with object detection. Lidar gives you a point cloud, and you need software that can detect and categorize things from that point cloud. Because lidar is new, this problem hasn't had as much R&D put into it as similar problems in computer vision.
Then there's the issue of sensor fusion. Lidar requires sensor fusion because it can't see road lines, traffic signals, or signs. It also can't see lights on vehicles or pedestrians. So you still have to solve most of the computer vision issues and you have to build software that can reliably merge the data from both sensors. What if the sensors disagree? If you err on the side of caution and brake if either sensor detects an object, you get lots of false positives and phantom braking (increasing the chance of rear-end collisions). If you YOLO it, you might hit something. If you improve the software to the point that lidar and cameras never disagree, well then what do you need lidar for?
I think lidar will become more prevalent, and I wouldn't be surprised if Tesla added it to their vehicles in the future. But the primary sensor will always be cameras.
Because LIDAR is expensive and available in limited quantities. There's no way they could sell the amount of cars they are selling right now if each one came with a LIDAR.
Camera modules are cheap and available in huge quantities.
Flies hit windows and cars. Even much more complex animals cannot manage effectively moving in a complex environment without hitting each other (like flocks of sheep).
There's no data that would support that anything than human level AGI is required to drive cars with how current infrastructure looks like.
To be fair, though, Tesla has no sensors other than cameras, and I believe the Mercedes has a half dozen or so, several radars and even a lidar.