Enhanced Summon, where are you?

[Daniel in SD]

Tesla has added a new "Stand by Mode" to Enhanced Summon in the Early Access version that went out to 2019.20.4.5.

“Summon Standby will keep the sensors, cameras, and computers powered to reduce the amount of time required to begin Summon."

https://electrek.co/2019/06/27/tesla-standby-mode-enhanced-summon-faster/

Saving the planet one kWh at a time
I wonder how much juice this sucks down.

 

[diplomat33]

Saving the planet one kWh at a time
I wonder how much juice this sucks down.

Well, we know Sentry Mode which also keeps sensors, cameras and computers on, drains about 1 mile/hr from the battery. So, I imagine "Summon Stand by" would be similar. If true, the battery drain would not be huge but would definitely be noticeable if you left it on for days at a time. I wonder if Tesla could add a button on the phone app to toggle "stand by" on and off. That way, we could toggle it on from our phone a few minutes before we plan to use enhanced summon rather than leave it on continuously.

 

[Daniel in SD]

Well, we know Sentry Mode which also keeps sensors, cameras and computers on, drains about 1 mile/hr from the battery. So, I imagine "Summon Stand by" would be similar. If true, the battery drain would not be huge but would definitely be noticeable if you left it on for days at a time. I wonder if Tesla could add a button on the phone app to toggle "stand by" on and off. That way, we could toggle it on from our phone a few minutes before we plan to use enhanced summon rather than leave it on continuously.

So if you uncheck those boxes:
290Wh/mi (EPA from wall) * 24 * 365 = 2540kWh per year which is close to the worldwide average electricity consumption per capita.

 

[AlanSubie4Life]

So if you uncheck those boxes:
290Wh/mi (EPA from wall) * 24 * 365 = 2540kWh per year which is close to the worldwide average electricity consumption per capita.

I've driven 18583 miles in one of my Spark EVs over the last 33 months, at 3.6 mi/kWh.

So, 18583mi / 3.6mi/kWh = 5162 kWh (battery to wheels). So say 1.15x that to account for charging losses, 6000kWh over 2.75 years.

That's 2100kWh per year for driving, vs. 2540kWh per year for keeping enhanced summon ready to go at all times.

So I guess we will both save lives and save the world by not using Enhanced Summon? Probably will save our sanity too.

Of course, my house's standby consumption (not counting things like the refrigerator) is about 250W these days too, in spite of my best efforts to reduce it, which is disgraceful (2200kWh/yr - as much as it takes to drive a car all year). Sadly my hands are tied on a lot of the culprits. I've been looking for a super-efficient doorbell using a switching regulator rather than a transformer for years but they don't seem to make one.

Anyway, the lesson here is to eliminate standby losses, and thereby save the world.

 

[dgatwood]

Tesla has added a new "Stand by Mode" to Enhanced Summon in the Early Access version that went out to 2019.20.4.5.

“Summon Standby will keep the sensors, cameras, and computers powered to reduce the amount of time required to begin Summon."

https://electrek.co/2019/06/27/tesla-standby-mode-enhanced-summon-faster/

This screams "hack" to me. I'd have expected their engineers to do this in a much more clever way.

To summon from any significant distance, you must use your phone anyway. So there's no point in keeping the hardware powered up except when the app is running and talking to the car. So Tesla should remove this user interface, and replace it with:

• An API call that powers up the hardware, along with support on the car side.
• Code in the app that makes that API call on launch or foregrounding.
• A timer that shuts down the hardware three minutes after the last time that the Summon feature moved the car.

Keeping the hardware hot all the time seems rather silly.

 

[AlanSubie4Life]

I can't imagine this will roll out to the masses. It seems completely silly; it doesn't take long to wake up the car from the app anyway as long as LTE connectivity is good. They were probably having some silly problem with there being a huge delay, and just wanted to get the feedback from the early access crowd sooner rather than later. As you say, a hack. To get around an inconsequential problem which probably won't be solved for another year (like the blank backup camera issue).

 

[rowdy]

Keeping the hardware hot all the time seems rather silly.

Probably required to detect objects going in to & out of the blindspots near the car.

Otherwise you're relying on waking up and not being fully aware of the surroundings (they may have changed while the car was asleep). The current hack of creeping slowly and relying on ultrasonics to fill in the blanks may not have proven 100% accurate in testing...

 

[AlanSubie4Life]

Probably required to detect objects going in to & out of the blindspots near the car.

But they SAY it is just to reduce the amount of time to begin Summon. They don't say you can't use Summon if you don't leave it in Standby Mode.

However, the blindspots around the car are going to be a curious problem to solve. We'll see!

 

[EVNow]

@Maitri982 @EVNow

I’m all for agreeing to disagree, but I do think you are naive if you don’t consider that marketing is likely a major motivation behind Elon’s tweets and that’s what I tried to get across. Yes, I do find it likely Elon’s overly optimistic timelines have a marketing motivation behind them.

I guess your response is understandable for people who see Tesla as a stock first and a car maker only second and I try not to take it as an ad hominem. For me Tesla is only a car maker and the only thing I’ve invested in them is buying a car from them. As for shorting the only thing I’ve ever shorted in my life are my clumsy electronics projects.

There was a purpose behind my shorting comment.

A pattern of behavior that is consistent with an assumption can be taken as proof for that assumption. For you all of Musk's tweets seem to be for generating demand. Similarly if I were to assume you are a short, all your comments are consistent with that assumption.

While Musk definitely uses his account to Market Tesla (and why not !) - not every tweet is designed that way. He has said it is a way of relating to his "fans" on twitter - its his way of relaxing. He is a celebrity, after all. BTW, he has also tweeted that if someone doesn't want to buy Tesla, they should buy EVs made by others. Can you find similar tweets by any other auto CEO ?

Elon Musk on Twitter

Or would someone only into marketing, tweet this ?

 

[electronblue]

There was a purpose behind my shorting comment.

A pattern of behavior that is consistent with an assumption can be taken as proof for that assumption. For you all of Musk's tweets seem to be for generating demand. Similarly if I were to assume you are a short, all your comments are consistent with that assumption.

While Musk definitely uses his account to Market Tesla (and why not !) - not every tweet is designed that way. He has said it is a way of relating to his "fans" on twitter - its his way of relaxing. He is a celebrity, after all. BTW, he has also tweeted that if someone doesn't want to buy Tesla, they should buy EVs made by others. Can you find similar tweets by any other auto CEO ?

Elon Musk on Twitter

View attachment 424164

Or would someone only into marketing, tweet this ?

View attachment 424165

None of us can get in the heads of others so all we of course ever have is our insight and interpretation.

First of all, Elon’s seeming promotion of other BEVs took a bit of a hit when he made fun of e-tron as étron on Twitter and not in a nice way. So, color me a bit skeptical as to how much of that is virtue signalling or not.

But of course Elon tweets about a range of topics for a range of reasons. I’d prefer not to share my cynical views on those for fear of derailing us further. However, my point was specific: I personally believe it likely he uses unrealisticially short timelines (not just in tweets but in other communications) for marketing purposes.

Put it another way. I think it is likely Elon realizes more realistic timelines would be bad for marketing so, the theory goes, he keeps sticking to unrealistic timelines.

Let’s take an example. In September 2016 Musk and Tesla gave a few timelines regarding AP1 upgrades that coincided with the end of Q3/2016 and Q4/2016 respectively (”8.0” and ”8.1”). Then he gave another timeline in October 2016 for AP2 that coincided with end of Q4/2016 (EAP delivered according to Tesla’s Design Studio, AP1 parity according to some other interpretations of his words).

The fact that neither of the Q4/2016 promises never materialized (AP1 navigation based exit taking in 8.1 and AP2 EAP or at least parity with AP1 eg speedsigns) is not really here nor there regarding my point. The point is what I believe the timelines were meant to communicate: Give people a reason to buy the cars that are available right now (both AP1/AP2 at that time, significant AP1 inventory/deliveries especially in the international channel still and also a sense that AP2 is already good enough to buy).

It is also not a surprise to me that Elon’s next comment on FSD features arriving have a timeline of only one to two quarters in January 2017. The message was once again clear, buy an FSD enabled Tesla now.

I mean rinse and repeat enough times and the likely parttern in my view has become clear. I think the guy is smart enough to know things likely take longer than that, especially by now. He could easily give rougher ballparks that would take this into consideration but he rarely seems to do that. I don’t buy it is to ”rally the troops to deliver” either, I think it is likely because he knows quick availability causes publicity in the media, which in turn causes interest in people and interest turns into purchases... much more so than saying ”well, this and that will happen two-three years from now” which actually would have been the more realistic thing to say in the AP2 case.

I mean, people forget: EAP with the equivalent Enhanced Summon (!!!) was promised for December 2016 according to the original Design Studio wording. And indeed originally ”meet you at the curb” Summon that turns the car around was already promised for AP1 in 2015.

Also, at least on Twitter, Elon almost never makes a distinction between releasing something and releasing something to Early Access or even just alpha testing. That too seems intentional. Get the publicity without really having to have something ready for a real launch. And it works too: Electrek just reported how Tesla released a new Standby Mode to Enhanced Summon. Unless you know what you are reading you’d be forgiven for thinking this is some kind of feature available to all Tesla cars and not in fact only a small group of confidential testers... That is nice publicity to have at the end of a quarter...

Heck, some might even think all old FSD owners actually got to Early Access Program because Elon tweeted it and much of the BEV media wrote about it. Never happened of course but it was good marketing.

 

[boonedocks]

• 
  
  So this thread started on 5-6-19 and today a new date has been set (approximately) by Elon....do you believe this latest prognostication?
  
  
  
  
  
  Elon Musk
  @elonmusk

Cost of Tesla full self-driving option increasing by ~$1000 on August 16
12:51 PM · Jul 16, 2019 · Twitter for iPhone

• Elon Musk
  @elonmusk

Replying to
@ropiko
Agreed. That’s approximately date when we expect Enhanced Summon to be in wide release. It will be magical. Lot of hard work by Autopilot team.
12:54 PM · Jul 16, 2019 · Twitter for iPhone

 

[EVNow]

Lot of hard work by Autopilot team.

My question has been - is the summon handled by a different team than FSD team ? Are there separate FSD and autopilot teams ?

In Karpathy presentation one thing stood out. He talked a little about summon (first thing they are doing outside highways) and said its great and added with a smile "when it works".

So, almost sounds like he is not in charge of that - and is done by a different team.

Comments ?

 

[RDoc]

In Karpathy presentation one thing stood out. He talked a little about summon (first thing they are doing outside highways) and said its great and added with a smile "when it works".

There was a little girl
Who had a little curl
Right in the middle of her forehead.
And when she was good,
She was very very good.
And when she was bad,
She was horrid.

 

[diplomat33]

• https://twitter.com/elonmusk
  
  So this thread started on 5-6-19 and today a new date has been set (approximately) by Elon....do you believe this latest prognostication?
  
  
  
  
  
  Elon Musk
  @elonmusk

Cost of Tesla full self-driving option increasing by ~$1000 on August 16
12:51 PM · Jul 16, 2019 · Twitter for iPhone

• Elon Musk
  @elonmusk

Replying to
@ropiko
Agreed. That’s approximately date when we expect Enhanced Summon to be in wide release. It will be magical. Lot of hard work by Autopilot team.
12:54 PM · Jul 16, 2019 · Twitter for iPhone

It kinda makes you wonder why Elon did not just give the Aug 16 deadline from the beginning?

 

[drtimhill]

So waiting for all the hate mail....

If they cant do advanced summon...full self driving is a pipe dream. I love my EAP, it solves 90% of all my driving...but the last 10% that fsd is supposed to address is a really hard nut to crack. You basically have to be able to account for all real world use cases:

1. Pedestrians jumping into traffic
2. Cars swerving into your lane and having to find exit routes in 100-500 milliseconds
3. trucks dropping items from their beds
4. one way streets becoming 2 way
5. lanes ending + merging
6. lane dividing paint being faint or non existent
7. Bad weather ..poor visibility

these are just some of the use cases....now take the 6 above and combine them...there could be cases where you have faint lane lines and pedistratians jumping onto the city street. Now try to imagine how complex this could get really fast.

I think they will deliver something better than EAP (90% functionality)...but how close to 99.9% remains to be seen.

But what should the benchmark be? Let’s assume (without prejudice!) that Tesla release an FSD that can be shown to have a 5x lower accident rate than humans. Is that good enough? Why not? it’s not some absolute value we are looking at, it’s “Is it as good as, or better than, human drivers” and lets face it THAT benchmark is pretty low!

But what will this future FSD driver profile look like?
— Almost no minor accidents, as the FSD has lightning-fast reactions compared to humans.
— Almost all accidents reduced in severity as car does the “sensible” thing rather than the (usually misguided) panic reactions of a human.
— Very occasional horrendous accident where for no reason the car does something fatal (drives off a cliff).

—Tim

 

[Daniel in SD]

But what should the benchmark be? Let’s assume (without prejudice!) that Tesla release an FSD that can be shown to have a 5x lower accident rate than humans. Is that good enough? Why not? it’s not some absolute value we are looking at, it’s “Is it as good as, or better than, human drivers” and lets face it THAT benchmark is pretty low!

Of course 5x is good enough. I think it would make sense to release FSD once it's only a little bit better than humans. I bet a big challenge in developing an autonomous vehicles is dealing with human drivers. Once autonomous vehicles are better than humans the roads will probably get much safer as they displace all the human drivers on the road.
People say humans are really bad at driving but machines are currently way worse. I read that Cruise vehicles have a mission failure (get stuck) every 30 miles and have a critical failure (driver intervention required to prevent an accident) every 3000 miles. Humans get in a serious accident every 300k miles.

 

[willow_hiller]

I think it would make sense to release FSD once it's only a little bit better than humans.

I was thinking about this earlier. It doesn't even need to be better than humans all the time. It just needs to be better than humans at some tasks some of the time.

As long as people don't use that slight comparative advantage as an excuse to completely ignore the road, human attentiveness + FSD will always be safer than just human attentiveness alone.

 

[Daniel in SD]

I was thinking about this earlier. It doesn't even need to be better than humans all the time. It just needs to be better than humans at some tasks some of the time.

As long as people don't use that slight comparative advantage as an excuse to completely ignore the road, human attentiveness + FSD will always be safer than just human attentiveness alone.

I was talking about actual Full Self Driving. I'm not convinced human assisted "Full Self Driving" will be safer. I guess it all depends how well Tesla can enforce attentiveness. Personally, I'd rather just drive myself in the city, monitoring FSD sounds way too stressful.

 

[willow_hiller]

I was talking about actual Full Self Driving. I'm not convinced human assisted "Full Self Driving" will be safer. I guess it all depends how well Tesla can enforce attentiveness. Personally, I'd rather just drive myself in the city, monitoring FSD sounds way too stressful.

I guess it's a good question what state Tesla is willing to release FSD in. It doesn't need to be un-monitored yet because no state would allow that immediately. But if it's in a buggy alpha state when released to the public, that could hurt Tesla's chances of getting the right regulations in place.

 

[EVNow]

I was talking about actual Full Self Driving. I'm not convinced human assisted "Full Self Driving" will be safer. I guess it all depends how well Tesla can enforce attentiveness. Personally, I'd rather just drive myself in the city, monitoring FSD sounds way too stressful.

Only way to figure out how good something is would be to release to public and collects stats. There will be people who use some things and others won't. That's fine.