Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

GM to introduce "Super Cruise" in its Cadillac CT-6 this fall

This site may earn commission on affiliate links.

bro1999

Active Member
Apr 26, 2016
2,793
3,574
Maryland
  • Like
Reactions: gowthamn
Based off the description, it sounds like you would be able to keep your hands off the wheel while Super Cruise is engaged without getting nagged by beeps and buzzes....as long as the system detects your eyes are pointed at the road ahead.
 
Based off the description, it sounds like you would be able to keep your hands off the wheel while Super Cruise is engaged without getting nagged by beeps and buzzes....as long as the system detects your eyes are pointed at the road ahead.

Obviously, nobody is going to know until it's released. Cadillac earlier said Dec.

One big hangup was the Cadillac's desire to be able to go into 100% autonomous mode briefly on public highways if the driver is not conscious. SuperCruise was slated for the 2016 CT6, but the NHTSA apparently stalled it, so they finally released the CT6 late without it. It is possible that's why the Bolt was late, and why it was released with no ACC.

So will it pull over safely at the nearest safe place and call for help? Or did they have to cancel that?
 
The hands off the wheel surprises me. What happens if there is construction? A crashed car? I closed lane for some reason?

Mapping the locations doesn't sound terribly dissimilar to Tesla's gathering of self driving data from each Tesla's driving, gathering data to better drive an area, but that doesn't help accommodate unexpected circumstances on the road.
 
So in case of construction, obstacles or some other issue not caught on the LIDAR maps its a one camera forward system with all the hype of "LIDAR"

Seems pretty half-baked for 2018

Even worse, the LIDAR-mapped high resolution maps mask the underlying performance for most users so they don't know how well (or badly) the system performs when meeting new threats. You'll actually have to go put a table on the freeway to see what happens during hands-off driving when the car encounters a situation it wasn't programmed for.

It's easy to just replay driving the same route over and over again with some level of safety. That's pretty much every inattentive driver commuting to and from work.
 
Even worse, the LIDAR-mapped high resolution maps mask the underlying performance for most users so they don't know how well (or badly) the system performs when meeting new threats. You'll actually have to go put a table on the freeway to see what happens during hands-off driving when the car encounters a situation it wasn't programmed for.

It's easy to just replay driving the same route over and over again with some level of safety. That's pretty much every inattentive driver commuting to and from work.

So in case of construction, obstacles or some other issue not caught on the LIDAR maps its a one camera forward system with all the hype of "LIDAR"

Seems pretty half-baked for 2018

Super Cruise
Multiple Cameras
Multiple Radars
Lidar CM(s) Accurate Map

Tesla
One Camera
One Forward Radar
GPS Logging Meter(s) Accurate Map


This makes no sense. Its uses more technology than tesla but all of a sudden now its a negative?
The reason it can pull you over the side of the road is because of its cm accurate map.
 
  • Disagree
  • Like
Reactions: Beryl and gowthamn
Super Cruise
Multiple Cameras
Multiple Radars
Lidar CM(s) Accurate Map

Tesla
One Camera
One Forward Radar
GPS Logging Meter(s) Accurate Map


This makes no sense. Its uses more technology than tesla but all of a sudden now its a negative?
The reason it can pull you over the side of the road is because of its cm accurate map.

From the description, this seems like less technology due to the difference in software. Relying on previously generated centimeter accurate maps seems like a very bad idea. I'd like to see it operate without those maps to make sure the underlying driving algorithms are decent enough.
 
From the description, this seems like less technology due to the difference in software. Relying on previously generated centimeter accurate maps seems like a very bad idea. I'd like to see it operate without those maps to make sure the underlying driving algorithms are decent enough.

No its rather your lack of understanding of the tech. It uses the map in addition to the lanes already detected using the cameras.
So the lanes and environment detected and the gps localize the car in the lidar map. Giving it precise detail of the environment its in.
Letting it work on hills, curve roads, allow it to pull over to the side of the highway. It knows its the side of the highway and not another random lane.

Its similar to the lidar map that google uses but for highway only.
 
As I said in the other thread, this tech is likely the next gen mobileye system.

It makes sense for GM to follow multiple development paths to autonomous driving, They probably didn't halt their mobileye development when they bought Cruise Automation.

The lidar mapping is interesting, and perhaps is a result of the limitations found in AP1.

Even worse, the LIDAR-mapped high resolution maps mask the underlying performance for most users so they don't know how well (or badly) the system performs when meeting new threats. You'll actually have to go put a table on the freeway to see what happens during hands-off driving when the car encounters a situation it wasn't programmed for.

It's easy to just replay driving the same route over and over again with some level of safety. That's pretty much every inattentive driver commuting to and from work.

That's not how it works. If the system very precisely knows "normal", it is much easier to sense exceptions.

Musk claimed a short cut to self driving in August of last year. How is that going? The system doesn't even detect stopped objects yet. How might Tesla get around that problem? By precisely mapping landmarks like buildings and bridges.
 
Last edited:
  • Like
Reactions: gowthamn
Yes indeed. If you had Lidar maps, you could rule out radar false positives for things like overhead signs. The same thing can be done with cameras (humans do it all the time), it just needs fancy image recognition algorithms, something Tesla is no doubt working on. The bottom line is that we don't know enough details about what GM or Tesla is doing to predict how well it will work. The proof is in the implementations they release. I can't wait to test it!
 
Maps are crucial to any self-driving system; you want them there - they give the car foresight. You need robust sensing AND high precision maps before you can get anywhere close to self-driving.

The map data is likely coming from TomTom or Here.