Driverless Cars' AI Drivers are Huge Power Hogs

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,719
Anyone that owns a high end GPU can vouch for large levels of power consumption. With that in mind, think about how power hungry all these new AI processors are. Now think about putting all those in an electric driverless car. It seems as though these AI driverless systems are going to put a significant ding in these car's fuel economy numbers. Makes sense to me. Autostart video warning over at Bloomberg.

LOL video.

Judging from General Motors Co.'s test cars and Elon Musk's predictions, the world is headed toward a future that's both driverless and all-electric. In reality, autonomy and battery power could end up being at odds.

That's because self-driving technology is a huge power drain. Some of today's prototypes for fully autonomous systems consume two to four kilowatts of electricity - the equivalent of having 50 to 100 laptops continuously running in the trunk, according to BorgWarner Inc. The supplier of vehicle propulsion systems expects the first autonomous cars - likely robotaxis that are constantly on the road - will be too energy-hungry to run on battery power alone.

"They're worried about one watt, and now you're adding a couple thousand," Thomas said. "It's not trivial."


Discussion
 
Heavens to Betsy, whatever you do don't turn on the Air Conditioning or Heating. ;p
Heating is the killer mostly. Electric A/C units are fairly efficient. This is just some of my own findings from my hybrid which isn't too much different than a full electric car.
Lights are now mostly LED which is a much more efficient than what they used to be (halogen).
 
high end gpu = 200-300 watt = 2-3 incandescent bulbs.
An electric motor uses on average 20-25k watts on a flat road.
so 200-300 vs 20000-25000 (on non challenging roads).
Not too sure where the concern is.
Self driving cars will require more than one GPU. 10 is not unrealistic at all, and 20 is quite probable when active standbys are accounted for. Plus the load for all the cameras, LIDARs, etc. and it starts to really add up.
 
That’s because self-driving technology is a huge power drain. Some of today’s prototypes for fully autonomous systems consume two to four kilowatts of electricity -- the equivalent of having 50 to 100 laptops continuously running in the trunk, according to BorgWarner Inc. The supplier of vehicle propulsion systems expects the first autonomous cars -- likely robotaxis that are constantly on the road -- will be too energy-hungry to run on battery power alone.

Huh? So the average laptop uses only 40 watts of power? I thought it was more like 60-100. Plus you put 50-100 laptops running in the trunk and youre going to end up with this:

car-fire.jpg


long before you run out of power...
 
Snippet:
A fully autonomous subcompact car like a Honda Fit, for example, will get 54.6 miles to the gallon in 2025 in the best-case scenario, more than 5 miles below the U.S. emissions target, according to BorgWarner. A small pickup or SUV would be at 45.8 mpg, versus a target of 50.

Im loathe to do it, but maybe we could lower the emission targets if these self-driving cars perform their promised miracle saving scenario of less cars on the road.
 
Has anyone evaluated the interaction/performance when you have a traffic jam full of all autonomous vehicles each pumping out their own flavor of radar, lidar, ultrasonic, wifi, and optical sensing?
 
This push to have all Electric is stupid. We need more power generation and the cars need much better batteries. A 5 year old ev range is going to be crap
 
Let's imagine it to be needing a new kind of processing technique our current CPU's has severe problems with, a little like when h264 and x265 was cropping up, then set-top boxes and the like needed a new chipdesign optimized for that.
Even if AI demands something extraordinary like that, then programmers will imagine what kind of extra features an "AI-chip" would need. And seeing demand for such designs, there will be optimized CPU's for this very tasks, and power-demand will drop.
Still, I see this as a small problem compared to living in Nordic climate zones and rather demanding winters. If electric cars cannot provide comfortable climate during winter months, then it is doomed in these regions at least. Then again I think cooling is actually even more power hungry in the Summer, on the other hand, it's easier to live with a somewhat too hot car than a freezing one, so it might differ from house climate system.
 
Snippet:


Im loathe to do it, but maybe we could lower the emission targets if these self-driving cars perform their promised miracle saving scenario of less cars on the road.

From my understanding, fuel efficiency should go up if every car was autonomous, even with the same number of cars on the road as now.
The primary cause for this, is that the autonomous cars are less likely to cause stop and go traffic or get stuck in 'invisible intersections'.
If vehicles are better able to maintain constant speed and not lose energy when slowing down, then mileage will go up.

That's not to say that 60mpg is still highly optimistic, considering the extra weight from safety features these days.
 
Fully autonomous Vehicles will require 10 to 15 of the newest Cray Super Computers and it would not be unfathomable if they required 20-30 more additional supercomputers for backup failsafes.
To power the above I'm guessing a small Thermal Nuclear Reactor similar to what's used on Air Craft Carriers, maybe 75 or 30 turbine engines like what is used on an M1 Abrams could be used for emergency backup power.
 
Heating is the killer mostly. Electric A/C units are fairly efficient. This is just some of my own findings from my hybrid which isn't too much different than a full electric car.
Lights are now mostly LED which is a much more efficient than what they used to be (halogen).

Heating?
/installs duct valve from CPU to car interior, FIXED!
 
Has anyone evaluated the interaction/performance when you have a traffic jam full of all autonomous vehicles each pumping out their own flavor of radar, lidar, ultrasonic, wifi, and optical sensing?

Forget traffic jams. What about the average downtown city? Signals from a few hundred cars bouncing off other cars, trucks, windows, buildings, hordes of people, etc.

If they do manage to solve the signal interference problem and the auto drive really does help with traffic flows, the savings in power from more efficient travel will likely make good the loss from powering the auto drive electronics.
 
It's inevitable that somebody's going to hack them and put them all to mine crypto, isn't it?
 
meh.

Auto manufactures need to demonstrate it works FIRST before worrying about the effect on fuel economy.
 
Back
Top