Autopilot power usage effect on range

kknogas

Active member
Joined
Jul 25, 2018
Messages
25
Location
Mobile, AL
Country
Country
Tesla Owner
Model 3
Has anyone done a test on the effect on the range when using autopilot on the highway vs not using it? In yesterday's autonomy investor day event, they showed a slide showing HW 2.5 using 57 watts and the new HW3 computer using 72 watts, and to one of the questions, Elon answered that it would reduce the consumer 250 watts per mile and 1 kWh per 4 miles. That is a decent impact on the range when using full self driving. Wonder what the impact of using autopilot currently is on the range.
 

MelindaV

☰ > 3
Moderator 📌
Joined
Apr 2, 2016
Messages
10,465
Location
Vancouver, WA
Country
Country
Tesla Owner
Model 3
Autopilot drives more efficiently than I do so it improves my range.
and I was going to say it is less efficient than my own driving. ;)

I am sure it varies depending on the type of driving, stop/go, constant speed, etc... but it does feel closer to my style of driving now than it did 4 or 5 months back in stop and go and moderate traffic - where it was very jerky and always going from full throttle to full braking previously. Now it is much better at reacting like a person.
 

RoBoRaT

Active member
Joined
Nov 24, 2018
Messages
48
Location
STUDENT AutoPilot in SoCalNorth
Country
Country
Tesla Owner
Model 3
Not done a test, but I notice about +10 mile left over in range on days that I use autopilot than not.

I go on AP on a mostly interstate 78 mile trip, one way.

In the morning at 4 AM, on AP 70-80 mph on interstate, 25/78 miles is cruising on HOV lane. Intervening once in a while. This trip is about 80 minutes.

On the way back, at 3 PM, AP but slower 60-70 mph on the 25 mile HOV lane, and the rest 40-60 mph, sometimes 0-40 mph stop and go traffic. I intervene about 3X more vs the AM trip. This trip averages 2 hrs.

I cannot explain why I end up with +10 mile left over range, but that is the trend I see when I use AP on this 160 mile drive.
 

jsmay311

Top-Contributor 👍🏻
Joined
Oct 2, 2017
Messages
466
Location
Chicago Suburbs
Country
Country
I did a test last year and noticed no difference.

It wasn’t the most scientifically rigorous test and it had a small sample size, but I drove an identical 2.5 mile route at 40mph 3 times in a row: one AP, one TACC, and one with neither. They all fell within 2 Wh/mi of each other.

Presumably FSD will consume more energy than today’s AP, but I think they mentioned at the autonomy event that the FSD chip will consume something like 72W(?). At a steady 40mph that’d still add less than 2 Wh/mi, and even less at higher speeds. (Pretty remarkable.)
 

kknogas

Active member
Joined
Jul 25, 2018
Messages
25
Location
Mobile, AL
Country
Country
Tesla Owner
Model 3
Not done a test, but I notice about +10 mile left over in range on days that I use autopilot than not.

I go on AP on a mostly interstate 78 mile trip, one way.

In the morning at 4 AM, on AP 70-80 mph on interstate, 25/78 miles is cruising on HOV lane. Intervening once in a while. This trip is about 80 minutes.

On the way back, at 3 PM, AP but slower 60-70 mph on the 25 mile HOV lane, and the rest 40-60 mph, sometimes 0-40 mph stop and go traffic. I intervene about 3X more vs the AM trip. This trip averages 2 hrs.

I cannot explain why I end up with +10 mile left over range, but that is the trend I see when I use AP on this 160 mile drive.

I'm not talking about the driving pattern of autopilot, but rather, the power usage of the autopilot computer and all the processing it needs to do when autopilot is activated.
 

kknogas

Active member
Joined
Jul 25, 2018
Messages
25
Location
Mobile, AL
Country
Country
Tesla Owner
Model 3
I did a test last year and noticed no difference.

It wasn’t the most scientifically rigorous test and it had a small sample size, but I drove an identical 2.5 mile route at 40mph 3 times in a row: one AP, one TACC, and one with neither. They all fell within 2 Wh/mi of each other.

Presumably FSD will consume more energy than today’s AP, but I think they mentioned at the autonomy event that the FSD chip will consume something like 72W(?). At a steady 40mph that’d still add less than 2 Wh/mi, and even less at higher speeds. (Pretty remarkable.)

Yes. They mentioned 72W usage doing FSD during the autonomy event and mentioned that it would be 250W per mile. Elon mentioned, the range is reduced more in city than on highway and said that if driving in city average 12mph, it would reduce the range by 25%. That's significant. That got me thinking how much farther we can travel without using autopilot vs using it currently.
 

slacker775

Top-Contributor 👍🏻
Joined
May 30, 2018
Messages
474
Location
Florida
Country
Country
Tesla Owner
Model 3
I wouldn’t be surprised if there is pretty close to zero effect on power usage because as they have indicated time and again, AP and its components are running in shadow mode if AP isn’t in actual usage. In other words, all of the same data is chugging through it, the only thing that isn’t happening is that it isn’t controlling the speed or steering wheel.

I’d bet you’ll notice after taking a decent length drive without even using AP at all that you’ll still upload a few MB of data up to the Tesla super computers in the sky as they collect events from the trip, or any scenarios that they are wanting to pull more detail on from the fleet.
 

Bokonon

Self-identified Teslaholic
TOO Supporting Member
Joined
Apr 12, 2017
Messages
3,186
Location
Boston
Country
Country
Tesla Owner
Model 3
Yes. They mentioned 72W usage doing FSD during the autonomy event and mentioned that it would be 250W per mile. Elon mentioned, the range is reduced more in city than on highway and said that if driving in city average 12mph, it would reduce the range by 25%. That's significant. That got me thinking how much farther we can travel without using autopilot vs using it currently.

Autopilot on the highway (Autosteer) uses only a fraction of the computing power than FSD requires, so I don't think that engaging it causes any meaningful increase in overhead power consumption. (I wish I could cite a specific number to illustrate just how much less power Autosteer uses vs. FSD, but do not have it handy.)

For FSD, even at 72W constant consumption, that only nets out to +6 Wh/mi at 12 mph average speed (72W / 12mi/h). For an AWD Model 3, that's only a 2.4% increase in energy consumption over the rated consumption of 250 Wh/mi, well within the realm of day-to-day fluctuations due to weather and driving conditions. And as jsmay311 pointed out above, the relative impact of that consumption drops significantly as average speed increases.

Maybe I'm missing something because I haven't yet watched the full FSD presentation and the Q&A, but I don't see present-day Autopilot or FSD consuming enough juice to impact range in any noticeable way. Unless your average speed was like < 3 mph. :)
 

kknogas

Active member
Joined
Jul 25, 2018
Messages
25
Location
Mobile, AL
Country
Country
Tesla Owner
Model 3
Autopilot on the highway (Autosteer) uses only a fraction of the computing power than FSD requires, so I don't think that engaging it causes any meaningful increase in overhead power consumption. (I wish I could cite a specific number to illustrate just how much less power Autosteer uses vs. FSD, but do not have it handy.)

For FSD, even at 72W constant consumption, that only nets out to +6 Wh/mi at 12 mph average speed (72W / 12mi/h). For an AWD Model 3, that's only a 2.4% increase in energy consumption over the rated consumption of 250 Wh/mi, well within the realm of day-to-day fluctuations due to weather and driving conditions. And as jsmay311 pointed out above, the relative impact of that consumption drops significantly as average speed increases.

Maybe I'm missing something because I haven't yet watched the full FSD presentation and the Q&A, but I don't see present-day Autopilot or FSD consuming enough juice to impact range in any noticeable way. Unless your average speed was like < 3 mph. :)



This is where Elon answers to a question about the range impact of FSD driving. He mentions losing 1 KW for 4 miles and range impact of 25%.
 

jsmay311

Top-Contributor 👍🏻
Joined
Oct 2, 2017
Messages
466
Location
Chicago Suburbs
Country
Country
Yes. They mentioned 72W usage doing FSD during the autonomy event and mentioned that it would be 250W per mile. Elon mentioned, the range is reduced more in city than on highway and said that if driving in city average 12mph, it would reduce the range by 25%. That's significant. That got me thinking how much farther we can travel without using autopilot vs using it currently.

This is where Elon answers to a question about the range impact of FSD driving. He mentions losing 1 KW for 4 miles and range impact of 25%.

I've transcribed the whole exchange below for clarity's sake.

(Note that Elon's "25% impact on range" comment is based on an explicitly hypothetical scenario in which a FSD system consumes 1000 Watts of power continuously -- a number he almost certainly chose just to make it easier to do the math in his head and not because it has anything to do with the actual expected power consumption of Tesla's FSD system.)

------------------------------------------------
Questioner: "On the power consumption, is there a way to give us a rule of thumb on, you know, every watt reduces range by a certain percent or a certain amount just so we can get a sense of how much..."

Pete Bannon: "On Model 3 the target consumption is 250 watts per mile[sic]*."

Elon Musk: "It depends on the nature of the driving as to how many miles that affects. In city it would have much a bigger effect than on highway. So if you're driving for an hour in a city and you had a solution, hypothetically, that was a kilowatt, you'd lose 4 miles on a Model 3.[**] So if you're only going , say, 12 mph, then that would be a 25% impact on range in city. The power to the system has a massive impact on city range, which is where we think most of the robotaxi market will be. So power is extremely important."
------------------------------------------------

*When first I heard that guy say "watts per mile" my head almost exploded. :rolleyes::p
"Watts per mile" is a nonsensical unit, of course, although it's often misused as shorthand for watt-hours per mile by non-engineers. So I believe here he's just talking about the rated battery-to-wheels energy efficiency of the Model 3 (albeit screwing up the units) in a clumsy attempt to put the 72W power consumption of the FSD chip into perspective. (I.e., that "250" number has nothing to do with the added energy consumption of the FSD hardware.)
 
Last edited:
  • Like
Reactions: 2 people

garsh

🐦🦆🦅🐧 🐓🦃🦉🦢🦜
Moderator 📌
TOO Supporting Member
Joined
Apr 4, 2016
Messages
15,850
Location
Pittsburgh PA
Country
Country
Tesla Owner
Model 3
I'm not talking about the driving pattern of autopilot, but rather, the power usage of the autopilot computer and all the processing it needs to do when autopilot is activated.
Even when you're not using autopilot, it's constantly running in "shadow" mode, and some differences detected between how you drive and what autopilot would have chosen are later uploaded to Tesla to investigate and possibly use to train a future version of the neural network.
 

Dr. J

Private
TOO Supporting Member
Joined
Sep 1, 2017
Messages
1,384
Location
Fort Worth
Country
Country
Tesla Owner
Model 3
Even when you're not using autopilot, it's constantly running in "shadow" mode, and some differences detected between how you drive and what autopilot would have chosen are later uploaded to Tesla to investigate and possibly use to train a future version of the neural network.
Crowdsourcing America's bad driving habits! :D
 

Ed Woodrick

Top-Contributor 👍🏻
Joined
May 25, 2018
Messages
1,313
Location
Atlanta, GA
Country
Country
Tesla Owner
Model 3
I believe the biggest impact on range is not the autopilot, but the speed that you are going and temperature.
When I was looking yesterday at my energy graphs, I was well under 250W for mixed driving and 80 F temps.
 

TomT

Well-known member
Joined
Apr 1, 2019
Messages
367
Location
NE Georgia
Country
Country
Tesla Owner
Model 3
I estimate that the difference in range is about that which is required to back out of the garage... :)
 
  • Like
Reactions: 1 person

kknogas

Active member
Joined
Jul 25, 2018
Messages
25
Location
Mobile, AL
Country
Country
Tesla Owner
Model 3
I've transcribed the whole exchange below for clarity's sake.

(Note that Elon's "25% impact on range" comment is based on an explicitly hypothetical scenario in which a FSD system consumes 1000 Watts of power continuously -- a number he almost certainly chose just to make it easier to do the math in his head and not because it has anything to do with the actual expected power consumption of Tesla's FSD system.)

------------------------------------------------
Questioner: "On the power consumption, is there a way to give us a rule of thumb on, you know, every watt reduces range by a certain percent or a certain amount just so we can get a sense of how much..."

Pete Bannon: "On Model 3 the target consumption is 250 watts per mile[sic]*."

Elon Musk: "It depends on the nature of the driving as to how many miles that affects. In city it would have much a bigger effect than on highway. So if you're driving for an hour in a city and you had a solution, hypothetically, that was a kilowatt, you'd lose 4 miles on a Model 3.[**] So if you're only going , say, 12 mph, then that would be a 25% impact on range in city. The power to the system has a massive impact on city range, which is where we think most of the robotaxi market will be. So power is extremely important."
------------------------------------------------

*When first I heard that guy say "watts per mile" my head almost exploded. :rolleyes::p
"Watts per mile" is a nonsensical unit, of course, although it's often misused as shorthand for watt-hours per mile by non-engineers. So I believe here he's just talking about the rated battery-to-wheels energy efficiency of the Model 3 (albeit screwing up the units) in a clumsy attempt to put the 72W power consumption of the FSD chip into perspective. (I.e., that "250" number has nothing to do with the added energy consumption of the FSD hardware.)

Thanks for transcribing it. Your explanation of what Elon said makes sense to me now. When I heard it, I assumed he was talking about losing 1 KW for 4 miles in the Model 3 with the FSD computer. But now, I understand he was talking about if some other computer uses 1000 watts, then the impact would be 25%. But since Tesla's FSD computer uses only 72W, it's going to be less than 2% compared to the 25% loss for a 1000W computer.
 
  • Like
Reactions: 1 person

PNWmisty

Legendary Member 🏆
Joined
Aug 19, 2017
Messages
2,684
Location
Anacortes, WA
Country
Country
But since Tesla's FSD computer uses only 72W, it's going to be less than 2% compared to the 25% loss for a 1000W computer.

Correct. Elon was really just highlighting the elegance of their FSD system vs. competitors whose software requires much more powerful computers.

If the computer draws 72W and you're traveling 72 mph, it's only adding 1 Wh/mile. As others have pointed out, the computer is running whether AP is engaged or not so that consumption is already baked into the assumed range.
 
  • Like
Reactions: 1 person