Summon seems like a silly party trick to me

[eye.surgeon]

Summon seems like a pointless party trick to me. 90% of the time it would be faster to just walk to your car in the parking lot than use summon. I wish Tesla would focus on genuine utility rather than party tricks. Perhaps those engineers could have spent that time on the production line improving initial quality? That would have been a much bigger ROI I suspect.

The kicker is that even if summon works 99% of the time, the 1% failures will be the ones in the headlines (this is already happening) so it's not even a bragging right for Tesla, more like a PR failure and also I suspect an insurance rate killer as there are going to be a bunch of $5,000 parking lot incidents in the next few months.

As is often the case with Elon there needs to be someone in the room when he shouts out ideas to say that's a stupid idea let's not waste development time on that.

 

[Fossil Fool]

There's the highly rational self that organizes pencils in the desk drawer and then there's the fun-loving irrational self that likes living on the bleeding edge of technology and being able to peer into the future. Sure, Summon is learning how to crawl, but it's Beta and down the road the experience will benefit other aspects of autonomy. If it was the final product there would be reason to complain.

I'm sure there will be headlines, but Tesla has survived worse. This technology allows the company to rightfully lay claim to the first production vehicle to have this ability.

 

[DirtyT3sla]

When AP 2 started it was limited to a much lower speed limit than 90...pretty much unusable. And yet here we are...these things take time.

 

[Tam]

...stupid idea...

I do agree that Smart Summon will be in the news for accidents.

At a maximum speed of 5MPH, the damages will be minimal (but still expensive).

I don't think that's a bad thing. I think that's informative as a reminder that it's still a beta system so don't fall asleep at the wheel!

I think Smart Summon is a great beta implementation for a potential driverless future.

The system learns how to avoid hitting obstacles including pedestrians. It learns stop sign, pedestrian crossing markings...

Those are basic skills that a driverless system needs to gain competency so it's best to learn in a controlled environment of 150 foot radius.

 

[boonedocks]

I have tried it in my small'ish very low traffic parking lot at my office and it failed both time and just stopped weirdly at an angle, like it just gave up. It may have continued after some time, I did not release the hold button for an addition 7-10 seconds with no continued movement, and gave up and ran to the car to get it out of the way.

Pretty simple scenario too:

 

[FunSecured]

I agree. I have been using V10 and Smart Summon since it came out and while it clearly has “miles to go,” it’s objectively amazing in what Tesla is continuing to do in moving from crawl to walk to run in autonomy.

Feels like some don’t necessarily (adequately?) regard the complexity of what is actually happening in this feature. Our cars are actively engaging in independent decision making, in real time.

Yes, it’s super slow still
Yes, it fails at times
Yes, it starts and sometimes stops
Yes, it’s range and the circumstances under which you can use or are pretty limited
All true

All that said, my experience is that when it gets it right, it’s just beyond cool, is something no other consumer car on the planet can do and really points to what’s gonna be another facet of the pathway forward in true autonomy.

This feature stands alongside Autopilot and Navigate on Autopilot and at least for me, I see how they’re all married and just going to keep improving and that this is beginning vs. end state.

In parking lots, for me, in more than one instance, Smart Summon has actually rightly recognized and stopped for pedestrians and vehicles and has rightly navigated to me despite raised curbs and multiple cars and requires turns. Though it was slow...it was amazing. Yes..I could have walked to my car quicker, but for me that really isn’t the “today” point in Smart Summon and it isn’t about it being “party trick” either. It’s about the current version being a “runway” start...like Auto Pilot and Navigate on Auto Pilot we’re and are.

 

[maverick3n1]

The other thing you forget is some people don't like to get soaking wet when it's pouring rain... it has some utility to it. Or perhaps you are carrying a heavy object you need to load up. It takes a lot less effort to have the vehicle come to you, then putting the package down, walking to the car, getting in and driving it to the package, parking and getting back out to load the package, then getting back in.

There are multiple uses for a feature like this, once it's perfected.

I think the main thing that frustrates me is not that it has problems, but that it has soooo many easily reproducible problems. Beta is for something that's been tested in-house and found to work, then released to the public for a larger scale test. This doesn't feel like it's in Beta, but instead in Alpha "Hey, we made this new feature but haven't really tested it. Lets release it and see what it does!".

When just about any of us can have a 50%+ fail rate of the feature, it feels like it was not tested at all. It's obvious that they just want to push out the product before it's ready to wow people, but as the OP stated, it means it's going to have a bunch of bad publicity when there are constant reports of fender benders. It also goes to show just how far away they are with self drive on city streets. They are supposed to be releasing that feature this year, and perhaps, because most city streets are google mapped, it will work a bit better as it has some GPS guidelines to follow vs. parking lots that aren't GPS mapped, but it still makes me scared to see what happens when they release that feature!

 

[davecolene0606]

Summon seems like a pointless party trick to me. 90% of the time it would be faster to just walk to your car in the parking lot than use summon. I wish Tesla would focus on genuine utility rather than party tricks. Perhaps those engineers could have spent that time on the production line improving initial quality? That would have been a much bigger ROI I suspect.

The kicker is that even if summon works 99% of the time, the 1% failures will be the ones in the headlines (this is already happening) so it's not even a bragging right for Tesla, more like a PR failure and also I suspect an insurance rate killer as there are going to be a bunch of $5,000 parking lot incidents in the next few months.

As is often the case with Elon there needs to be someone in the room when he shouts out ideas to say that's a stupid idea let's not waste development time on that.

So, when my daughter gets off of work late, she can stand safely next to the front door instead of walking alone to her car in a dark parking lot...

Priceless.

Fire Away!

 

[am_dmd]

The other thing you forget is some people don't like to get soaking wet when it's pouring rain... it has some utility to it. Or perhaps you are carrying a heavy object you need to load up. It takes a lot less effort to have the vehicle come to you, then putting the package down, walking to the car, getting in and driving it to the package, parking and getting back out to load the package, then getting back in.

There are multiple uses for a feature like this, once it's perfected.

I think the main thing that frustrates me is not that it has problems, but that it has soooo many easily reproducible problems. Beta is for something that's been tested in-house and found to work, then released to the public for a larger scale test. This doesn't feel like it's in Beta, but instead in Alpha "Hey, we made this new feature but haven't really tested it. Lets release it and see what it does!".

When just about any of us can have a 50%+ fail rate of the feature, it feels like it was not tested at all. It's obvious that they just want to push out the product before it's ready to wow people, but as the OP stated, it means it's going to have a bunch of bad publicity when there are constant reports of fender benders. It also goes to show just how far away they are with self drive on city streets. They are supposed to be releasing that feature this year, and perhaps, because most city streets are google mapped, it will work a bit better as it has some GPS guidelines to follow vs. parking lots that aren't GPS mapped, but it still makes me scared to see what happens when they release that feature!

If Tesla does not have problem with releasing AP (autosteer, NOA, wipers etc.) then why do you think they have problem with releasing alpha quality SS? Infact entire car is Beta at best. Tell me one thing hardware or software wise that's been properly tested

 

[Tam]

...This doesn't feel like it's in Beta, but instead in Alpha "Hey, we made this new feature but haven't really tested it. Lets release it and see what it does!"...

Waiting for a better beta is just too long.

It was promised 3 years ago in 2016.

It has been promised coming out soon "next week" such as promised again in April of this year.

It's been almost 6 months since April, so, I don't mind a very bad beta right now because I don't want to wait for it after 3 long years!

 

[diplomat33]

Summon seems like a pointless party trick to me. 90% of the time it would be faster to just walk to your car in the parking lot than use summon. I wish Tesla would focus on genuine utility rather than party tricks. Perhaps those engineers could have spent that time on the production line improving initial quality? That would have been a much bigger ROI I suspect.

You are forgetting that there are technologies that went into Smart Summon that will undoubtedly be useful for other features and FSD in general, for example, camera detection of curbs. So the work on Smart Summon still contributes to FSD even if the current Smart Summon does not seem all that practical to you. Plus, Smart Summon will improve. So the work that went into Smart Summon will help make it better in the future.

 

[am_dmd]

So, when my daughter gets off of work late, she can stand safely next to the front door instead of walking alone to her car in a dark parking lot...

Priceless.

Fire Away!

Or she could get stuck with a damaged car that may or may not be drivable
Of course I don't wish this upon her or anyone but that's the risk.

 

[Puddles]

Silly party trick?

I'll take a dozen!

 

[SandiaGrunt]

there are technologies that went into Smart Summon that will undoubtedly be useful for other features and FSD in general, for example, camera detection of curbs

See the camera detection of curbs in action at about 4:00.

It's going to be truly delicious, @diplomat33, when you finally have to admit that your optimism has been unwarranted.

 

[diplomat33]

It's going to be truly delicious, @diplomat33, when you finally have to admit that your optimism has been unwarranted.

Why are you taking pleasure in hoping someone loses their optimism? Frankly, that's pretty twisted.

I readily admit that Smart Summon is imperfect. The reason I am optimistic is because I want Smart Summon to get better in the future. Don't you want that too?

 

[SandiaGrunt]

Frankly, that's pretty twisted.

Is it more or less twisted than testing shitty, half-baked, potentially deadly software in parking lots full of children?

 

[diplomat33]

Is it more or less twisted than testing shitty, half-baked, potentially deadly software in parking lots full of children?

That is an grossly unfair and hyperbolic mischaracterization of Smart Summon.

 

[electronblue]

You are forgetting that there are technologies that went into Smart Summon that will undoubtedly be useful for other features and FSD in general, for example, camera detection of curbs. So the work on Smart Summon still contributes to FSD even if the current Smart Summon does not seem all that practical to you. Plus, Smart Summon will improve. So the work that went into Smart Summon will help make it better in the future.

I doubt curb detection went into Smart Summon, it seems like a generic part of the ”Tesla Vision” NNs. They will need to do that anyway. However, it is interesting that curb detection comes up again. We have discussed it before and I’ve made the point the major blindspot around the forward quarter of the car makes it susceptible to exaclty the kind of curb-grazing issues witnessed on the video in this thread. The car simply can not see the curbs around its nose up close, there are no sensors to detect them.

Why are you taking pleasure in hoping someone loses their optimism? Frankly, that's pretty twisted.

I would like us to improve our understanding of Autonomous Vehicles. So far I am of the opinion your optimism isn’t often warranted and thus it might be fruitful for all if it came down a notch. Just in the name of keeping things real.

I may be wrong. But just my honest opinion.

 

[SandiaGrunt]

That is an grossly unfair and hyperbolic mischaracterization of Smart Summon.

I have a beautiful, bubbly, happy 14-month old toddler who walks better than some 2 year olds. I watch all these Smart Summon videos with rednecks literally saying "I dunno, I dunno, I can't see it from this angle, I think it's gonna come around the long way."

I think about the very real possibility that one day, some goddamn zombie Tesla, "piloted" by someone I can't even see, is going to turn a corner and hit my child. They won't even know it happened, and will just keep holding that button down with their fat thumb.

This is going to happen to someone, probably very soon.

There is evidence all over TMC that the cars can't properly detect curbs, trash cans, garage walls, grass, reversing vehicle, stop signs, parking space lines, or semi trucks. I have literally zero faith that they can detect prone toddlers.

 

[SandiaGrunt]

That is an grossly unfair and hyperbolic mischaracterization of Smart Summon.

Here's how I described Smart Summon: "shitty, half-baked, potentially deadly software in parking lots full of children."

Which parts do you feel are "grossly unfair" or "hyperbolic?"

Parking lots are inarguably full of children.

AV software is inarguably potentially deadly.

The consensus among the media, and even among Tesla fans right here in this thread, is that Smart Summon is half-baked.

So.... which part of my statement is "grossly unfair" or "hyperbolic," @diplomat33? Is it just the word "shitty?" Are you so thin-skinned?