Tesla to launch Full Self Driving beta to select drivers next week

[Artiste]

So many armchair cynics... What a great British passtime to ctitisise someone who is doing something from the comfoort of your computer chair.

Not here to defend Elon, but how does "a beta of full FSD rerwrite to be released" so we can see new functionality be turned from exciting to "wake me up when it can pick me up from Heathrow" and "Elon musk is full of BS" thread?

I get that it can be fun to be so negative, but give it a rest already, 2020 has enough negativity without turning positive news into negative news.

Cynicism or reality? Negative or going on past experience? It’s great to be optimistic, but also important not to be carried away by hype. I don’t think there’s anything wrong in calling Elon, or anyone, out when they make such outrageously inaccurate statements.

 

[Yev000]

Cynicism or reality? Negative or going on past experience? It’s great to be optimistic, but also important not to be carried away by hype. I don’t think there’s anything wrong in calling Elon, or anyone, out when they make such outrageously inaccurate statements.

I don't think you are getting my point. Go ahead and rant about how bad FSD is, I've done plenty of that myself.

My point is, this news is positive, there is no need to "call it out", or "douse the hype". Noone is getting cheated out of their hard earned money here.

Also, its OK to be positive sometimes... Really, try it...

 

[Artiste]

I don't think you are getting my point. Go ahead and rant about how bad FSD is, I've done plenty of that myself.

My point is, this news is positive, there is no need to "call it out", or "douse the hype". Noone is getting cheated out of their hard earned money here.

Also, its OK to be positive sometimes... Really, try it...

I am an eternal optimist but I’m also thoroughly realistic and not taken in by hype. Then if the predictions turn out to be full of hot air I’m not disappointed. Some would call that cynicism, but being a positive bloke I call it realism.

 

[MrBadger]

It's negativity bourne of experience of the current iteration of FSD and how many problems it has though. Its not baseless

But it is baseless if you are comparing capabilities in 3 months time with its current capabilities. EM has said that the current software architecture is problematical and the new one will solve many (he probably said ALL but it is EM) of these.

So you need to start a fresh knowing that Tesla should have come up with a software architecture that addresses the shortcomings of the previous ones. No one publicly, except EM, has given any indication on how the new architecture performs, so until that happens (hopefully starting next week), its pure guess work.

Tesla do seem to have a top team, although they do seem to have wft moments. I must admit that I am surprised at how long it has taken them to feel the need to create a coherent view of the environment in both time and space. I guess it shows the calibre of the implementation of their current approach even if that approach is flawed.

But I think we need to give them time before judgement to be made. Currently, no evidence that FSD is anything other than software and regulatory limited. When rewrite is complete and mature, they will have nowhere to hide though, except maybe regulations. But there is much work being done on those, but perhaps not at a speed of everyone's liking and I suspect may also need a change of approach of rollout if Tesla are going to keep regulators happy.

 

[Yev000]

I am an eternal optimist but I’m also thoroughly realistic and not taken in by hype

You bought a Tesla... You are on a Tesla forum....

You are on the hype train...

You sound bitter about it.

 

[Drew57]

Past performance is often an indicator of what will happen in the future & on that basis many of us are rightly cautious. However this is a potentially exciting development.

My attitude is very positive towards the car & I will closely follow these FSD developments as I may add this if it becomes available as a subscription and can be used safely (won't keep this model long enough to justify £6,800 or more)

Most of us probably have expectations that are realistic rather than negative & I genuinely wish Tesla well with this.

 

[Neilio]

But it is baseless if you are comparing capabilities in 3 months time with its current capabilities. EM has said that the current software architecture is problematical and the new one will solve many (he probably said ALL but it is EM) of these.

So you need to start a fresh knowing that Tesla should have come up with a software architecture that addresses the shortcomings of the previous ones. No one publicly, except EM, has given any indication on how the new architecture performs, so until that happens (hopefully starting next week), its pure guess work.

Tesla do seem to have a top team, although they do seem to have wft moments. I must admit that I am surprised at how long it has taken them to feel the need to create a coherent view of the environment in both time and space. I guess it shows the calibre of the implementation of their current approach even if that approach is flawed.

But I think we need to give them time before judgement to be made. Currently, no evidence that FSD is anything other than software and regulatory limited. When rewrite is complete and mature, they will have nowhere to hide though, except maybe regulations. But there is much work being done on those, but perhaps not at a speed of everyone's liking and I suspect may also need a change of approach of rollout if Tesla are going to keep regulators happy.

Your defence of Tesla is touching but the sad thing is there is no evidence at all that a software that can't yet work properly on easy motorway drives will miraculously be able to do all driving in 3 months. Thats my point. No software re write can cover the fact we are so far off FSD, be a significant amount. It can't cope with rain, or lorries, or bridges on an arrow straight road at the moment so thinking it'll cope with roundabouts and cars parked on the road in 3 months is a huge, massive stretch.

This is a company that can't yet write rain detecting software that detects rain properly yet.

 

[phil4]

Perhaps the issue is more that it's being called FSD. If it were SD then we're simply going from how well it works now to better. But instead it's the "Full" bit that means they're aiming for and claiming perfection, and we're not seeing it based on now.

Put another way, we've talked about the levels of autonomy, and perhaps Tesla should be aiming for a lower step, eg. self driving in certain conditions, rather than just level 5, FSD.

Would I buy it if it could manage 98% of my journey by itself (ie. the middle bit), quite likely - almost certainly if we can ditch the holding the steering wheel every few seconds. Would it be nice if it was 100%? Sure.

The other aspect to consider on the above is if it can manage 98% of my daily journey by itself, it'd be good to clearly know if that 98% is concrete, and the other 2% not. Ie. not having to "learn" and "trust" which bits it can get right or wrong. At the moment it's a bit like getting in a taxi, but you don't know when the driving might suddenly not be there, and need to be ready to take over at any moment... that's really not a great taxi journey to do.

 

[Hurricane]

I’m certainly am optimist and am very much looking forward to the software rewrite. Especially as I think FSD in its current form is a complete waste of money and I regret buying it. I was certainly taken in by the hype. Tesla are also, apparently, incapable of fixing my crap auto headlights and crap windscreen wipers.

I don’t think it’s unreasonable to be cautious though. Elon undoubtedly has a way of hyping things far beyond reality, and his timelines are even more inaccurate than one of Donald Trump’s campaign promises.

I don’t see any sign of the million robo taxis that are supposed to be on the road by the end of this year.

 

[rincewind]

I drove my son to school this morning. It's two and a half miles, some down b roads, some through a small town. In my view in that short journey there were at least four times when I believe fsd would have brought the car to a halt unnecessarily, whereas I would have seen the gap to go through and piloted the car accordingly. there were also other times when I don't believe it would have known what to do. There is a set of traffic lights that, even for humans, are really confusing. I can't see how the car would know which one was the right one to choose when it turned green. I think fsd is a long long way away.

 

[MrBadger]

Your defence of Tesla is touching but the sad thing is there is no evidence at all that a software that can't yet work properly on easy motorway drives will miraculously be able to do all driving in 3 months. Thats my point.

No one is saying that it can do all driving in 3 months. I think the spec is automatic driving on city streets (no operational domain specified) with no time frame other than upcoming?

You cannot compare the software as it is now and the software in 3 months time (or how ever long it take to mature). Its a complete architectural rewrite. What you will be using when the rewrite matures will be completely different software. Its not a refactoring exercise of code or network written for older hardware, but a fundamental architectural change written for current hardware. Past behaviour is not indicative of future behaviour, a Jaffa Orange is not indicative of the taste of a Terrys Chocolate Orange etc.

I think the issue is the 'Full' from the term Full Self Drive. Autonomy/self starts at Level 3. I think we are pretty close to Level 3 and/or Level 4 for motorways, regulations omitted. Agree much further away for city driving, but thats not even advertised as a current feature.

 

[GRiLLA]

I don't think it's too far fetched to assume that FSD will work by the end of the year/early next for early beta testers in some areas of America (regulations pending). Over here, and in most of europe is a different matter, many main roads and those between houses were built to allow a horse to get down not a car, so they've been adapted to allow cars. So there's a lot of good roads, and a LOT of roads way too narrow, with no road markings and multiple obstacles and potholes. While Elon says the best way for FSD is to not have pre planned maps of how to drive and to just learn to drive properly, there's a lot of places in the UK where even a competent driver would be confused as to how to navigate, so it would need to form of assistance and have learned data specific to that road. Not really an issue to do but it'll just be further down the line....

1/3 of all the roads in the US are unpaved, i.e. dirt roads, so all countries have places that are more challenging. I'm still going to stick to believing that AP only needs to be as smart as average person in the UK with a driving license, and that's really not a very high bar.

 

[Spacep0d]

Weirdly, I have zero interest in this.

 

[rincewind]

Weirdly, I have zero interest in this.

Nor I

 

[GRiLLA]

I appreciate that many people wrestled with the decision around buying FSD or not, and now really want to believe they made the right choice. Their confirmation bias will latch on to any rationale that backs up their decision.

It must be tiring being negative all the time, much more fun to be optimistic. We're all deluded, none of us have a clue about this highly complex topic. Maybe it's polite to not hijack every thread to unload negativity.

 

[Latibes]

This car is light years ahead of anything currently on the road. Enjoy the car and its features!

 

[Glan gluaisne]

Weirdly, I have zero interest in this.

Same here.

I didn't buy the car because there was some vague hope that, at some unknown time in the future, it might be able to become self-driving to some degree. I bought it because it happened to meet our requirements for a car, plus (and this bit was probably the most significant) it was great fun to drive.

I'm sure Full Self Driving is important to some, and that's fine, but I suspect that, for many, it's just another nice-to-have feature, like having sat nav or phone integration.

I remain convinced that the most successful fully autonomous, self-driving cars will probably be purpose designed, rather than just a feature added to an otherwise conventional car. Apart from anything else, why would an autonomous car need to look anything like a conventional car? Moving away from the appearance and constraints imposed by having to have things like pedals, driver controls, a steering wheel, and even a large windscreen, opens up opportunities to make self-driving cars safer as well as more efficient.

Just doing as RAF Transport Command did for years, and having the seats facing backwards, would make autonomous cars safer in accidents, not that I think people would like it very much. It's an example of the way that moving away from conventional thinking could bring greater safety benefits.

 

[Plagued]

You cannot compare the software as it is now and the software in 3 months time (or how ever long it take to mature). Its a complete architectural rewrite. What you will be using when the rewrite matures will be completely different software. Its not a refactoring exercise of code or network written for older hardware, but a fundamental architectural change written for current hardware. Past behaviour is not indicative of future behaviour, a Jaffa Orange is not indicative of the taste of a Terrys Chocolate Orange etc.

This!
work on the current "autopiolot" pretty much stopped a year ago, there's been a bit of bleed through with teaching it to detect new objects like lights and bins, but the re-write while using the same hardware, is fundamentally a different system.
Reserve any judgement as to how good "autopilot" will be (good or bad) until we've actually seen it running. Right now other than Elon and a few people within Tesla, no one has actually seen how it performs.

 

[Type2]

This car is light years ahead of anything currently on the road. Enjoy the car and its features!

In some ways it most certainly is. In others it’s decades behind.

 

[Hurricane]

I appreciate that many people wrestled with the decision around buying FSD or not, and now really want to believe they made the right choice. Their confirmation bias will latch on to any rationale that backs up their decision.

It must be tiring being negative all the time, much more fun to be optimistic. We're all deluded, none of us have a clue about this highly complex topic. Maybe it's polite to not hijack every thread to unload negativity.

Let’s not confuse negativity with valid criticism. Criticism can bring improvements. If everyone on this forum said “oh wow, my windscreen wipers are the best thing I’ve ever experienced” then they would never improve.

Mind you, people have been criticising the windscreen wipers for years . . . and they are still crap