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

Waymo

This site may earn commission on affiliate links.
A true FSD does not need localization actually. A true FSD should and will have to evaluate actual conditions as they exist right then and there and not what was scanned a few days or months ago. It would need to be intelligent to keep itself safe while continuing it’s journey to its destination, whether or not lane markers, repeaters, traffic lights, signals, signs are available or in a usable state.

What? That's not what localization means. Localization means knowing where you are in relation to static features on the ground. True FSD absolutely needs precise localization because you can't make turns or even do lane keeping if you don't know your precise location. For example, if the AV needs to make a left turn at an intersection, it needs to know it's position in the intersection in order to calculate how to make the left turn. If the AV has bad localization, ie it thinks it 10 ft to the left or 10 ft forward than it really is, then it will have trouble making the left turn. Localization is also key for lane changes and navigation because if the AV thinks it is in the left lane when it is actually in the right lane, that changes a lot. So accurate localization is critical.
 
Last edited:
What? That's not what localization means. Localization means knowing where you are in relation to static features on the ground. True FSD absolutely needs precise localization because you can't make turns or even do lane keeping if you don't know your precise location. For example, if the AV needs to make a left turn at an intersection, it needs to know it's position in the intersection in order to calculate how to make the left turn. If the AV has bad localization, ie it thinks it 10 ft to the left or 10 ft forward than it really is, then it will have trouble making the left turn. Localization is also key for lane changes and navigation because if the AV thinks it is in the left lane when it is actually in the right lane, that changes a lot. So accurate localization is critical.
Then enlighten me on what is this “scan the city and build HD maps” prior to launch is all about.

Localization you described needs to happen in real time, not based on previous scans of that route.

My contention is that NO SCAN or HD Maps is needed for true FSD. Until that time it is going to be BSFSD.
 
Last edited:
  • Funny
Reactions: powertoold
Then enlighten me on what is this “scan the city and build HD maps” prior to launch is all about.

Localization you described needs to happen in real time, not based on previous scans of that route.

My contention is that NO SCAN or HD Maps is needed for true FSD. Until that time it is going to be BSFSD.
I'm imagining you telling the Wright Brothers that true flight can only be achieved by the flapping of wings. haha.
 
Then enlighten me on what is this “scan the city and build HD maps” prior to launch is all about.

Localization you described needs to happen in real time, not based on previous scans of that route.

My contention is that NO SCAN or HD Maps is needed for true FSD. Until that time it is going to be BSFSD.
"True FSD" appears to be a term you devised. As such, you are free to define it however you like.
 
Then enlighten me on what is this “scan the city and build HD maps” prior to launch is all about.

Waymo builds a HD map first to give the car a prior about the road and traffic rules. For example, HD maps allow the car to know what is up ahead, beyond visual range and see the road when it is occluded. HD maps also allow the car to know about local driving rules. The Waymo can check the HD map but it relies on cameras, radar and lidar to do the actual driving. So Waymo does not need HD maps to drive. It can drive based on just cameras, lidar and radar. But HD maps have helpful information and improve safety.

My contention is that NO SCAN or HD Maps is needed for true FSD. Until that time it is going to be BSFSD.

Your contention is wrong. HD maps improve safety. Without HD maps, the car would have to just figure stuff out again and again and guess what it can't see which would not be reliable. With HD maps, you will have higher reliability which is needed for true FSD, ie driverless.

Localization you described needs to happen in real time, not based on previous scans of that route.

Localization does happen in real-time. Like I said above, localization uses accelerometers, gyroscopes, wheels speeds, as well as lidar and radar to make real-time measurements of the vehicle's speed and heading. You seem to be confusing localization with mapping. Those are different things.
 
  • Like
Reactions: Bladerskb
Got an email asking me to join waymo. I looked at waymo website and they have a long list of job openings.
I found it interesting, thought I'd share.
 
Waymo builds a HD map first to give the car a prior about the road and traffic rules. For example, HD maps allow the car to know what is up ahead, beyond visual range and see the road when it is occluded. HD maps also allow the car to know about local driving rules. The Waymo can check the HD map but it relies on cameras, radar and lidar to do the actual driving. So Waymo does not need HD maps to drive. It can drive based on just cameras, lidar and radar. But HD maps have helpful information and improve safety.



Your contention is wrong. HD maps improve safety. Without HD maps, the car would have to just figure stuff out again and again and guess what it can't see which would not be reliable. With HD maps, you will have higher reliability which is needed for true FSD, ie driverless.



Localization does happen in real-time. Like I said above, localization uses accelerometers, gyroscopes, wheels speeds, as well as lidar and radar to make real-time measurements of the vehicle's speed and heading. You seem to be confusing localization with mapping. Those are different things.
HD maps is like the WRC Rally team reviewing exactly how the rally route is, and how and when to handle them. I get the intention you mean. But then, the roads are kept empty for the WRC Rally team. That is not the case for daily driving FSD. Anything and everything is subject to change on the route. So the HD maps in theory are good but are pretty irrelevant in actual application.

Eg. Tree fell down in the main road next to my house. 2 lanes are blocked. Construction is always ongoing and various lanes and exits are opened and blocked off as the metroplex grows bigger and bigger. A scan is as good as the time it was scanned.
 
So the HD maps in theory are good but are pretty irrelevant in actual application.
Yet there are driverless vehicles using HD maps so they also work in practice. I agree that HD maps are only a small part of their solution.
Eg. Tree fell down in the main road next to my house. 2 lanes are blocked. Construction is always ongoing and various lanes and exits are opened and blocked off as the metroplex grows bigger and bigger. A scan is as good as the time it was scanned.
This is a very black and white way of thinking. An AV does not need to perform better than a human in every possible situation to perform better than a human overall.
 
HD maps is like the WRC Rally team reviewing exactly how the rally route is, and how and when to handle them. I get the intention you mean. But then, the roads are kept empty for the WRC Rally team. That is not the case for daily driving FSD. Anything and everything is subject to change on the route. So the HD maps in theory are good but are pretty irrelevant in actual application.

Eg. Tree fell down in the main road next to my house. 2 lanes are blocked. Construction is always ongoing and various lanes and exits are opened and blocked off as the metroplex grows bigger and bigger. A scan is as good as the time it was scanned.

No, I don't think you get it. The car does not just drive based on what the HD map tells it to do. HD maps are not irrelevant because they provide useful information about the static world. But the car uses the cameras, lidar and radar to drive in real-time. So the car detects everything with its cameras, lidar and radar and determines the path to drive. Using your examples, the HD map would tell the car where the lanes should be but the car would detect the construction zones or the fallen tree with its cameras, lidar and radar and override the HD map. The car would plot a safe path in the construction zone or around the fallen tree using its cameras, lidar and radar. And keep in mind the cameras, lidar and radar detect all moving objects, like other vehicles, cyclists, pedestrians, objects on the road etc... So the car uses its cameras, radar and lidar to navigate safely. The HD map is only for static objects, not moving objects.

Here are a few examples:
  • AV is driving. HD map matches the lane lines that the cameras, lidar and radar see. Cool. The car continues driving with increased confidence since both sources match.
  • The car can't see a stop sign because a storm knocked it down the day before but the car knows there is a stop sign there because of the HD map and slows down and stops.
  • There is a dense fog. The cameras and lidar have reduced range and can't see the lane lines very far ahead. But the HD map tells the car where the lanes should be, far ahead. The car can still drive despite the cameras and lidar not seeing as far.
  • AV is driving. HD map tells it that there is a sharp bend coming up soon that the sensors can't see. So the AV knows to slow down in anticipation of the bend in the road.
  • Traffic is congested. Cameras, lidar can't see all the lane lines because all the vehicles. HD map tells the car where the lanes are.
  • Car comes up to a construction zone. Cameras, lidar and radar detect the construction zone. The car overrides the HD map. The car navigates the construction zone based on the cameras, lidar and radar.
  • Cameras, lidar, radar detect a fallen tree. The car plots a safe path around the fallen tree using the cameras, lidar and radar.

So as you can see: the HD map can help the car in some instances but the car also knows when not to use the HD map.
 
Last edited:
Yet there are driverless vehicles using HD maps so they also work in practice. I agree that HD maps are only a small part of their solution.

This is a very black and white way of thinking. An AV does not need to perform better than a human in every possible situation to perform better than a human overall.
AV needs to perform as good as a human. A human driver scans the environment and makes decisions in real time.
 
No, I don't think you get it. The car does not just drive based on what the HD map tells it to do. HD maps are not irrelevant because they provide useful information about the static world. But the car uses the cameras, lidar and radar to drive in real-time. So the car detects everything with its cameras, lidar and radar and determines the path to drive. Using your examples, the HD map would tell the car where the lanes should be but the car would detect the construction zones or the fallen tree with its cameras, lidar and radar and override the HD map. The car would plot a safe path in the construction zone or around the fallen tree using its cameras, lidar and radar. And keep in mind the cameras, lidar and radar detect all moving objects, like other vehicles, cyclists, pedestrians, objects on the road etc... So the car uses its cameras, radar and lidar to navigate safely. The HD map is only for static objects, not moving objects.

Here are a few examples:
  • AV is driving. HD map matches the lane lines that the cameras, lidar and radar see. Cool. The car continues driving with increased confidence since both sources match.
  • The car can't see a stop sign because a storm knocked it down the day before but the car knows there is a stop sign there because of the HD map and slows down and stops.
  • There is a dense fog. The cameras and lidar have reduced range and can't see the lane lines very far ahead. But the HD map tells the car where the lanes should be, far ahead. The car can still drive despite the cameras and lidar not seeing as far.
  • AV is driving. HD map tells it that there is a sharp bend coming up soon that the sensors can't see. So the AV knows to slow down in anticipation of the bend in the road.
  • Traffic is congested. Cameras, lidar can't see all the lane lines because all the vehicles. HD map tells the car where the lanes are.
  • Car comes up to a construction zone. Cameras, lidar and radar detect the construction zone. The car overrides the HD map. The car navigates the construction zone based on the cameras, lidar and radar.
  • Cameras, lidar, radar detect a fallen tree. The car plots a safe path around the fallen tree using the cameras, lidar and radar.

So as you can see: the HD map can help the car in some instances but the car also knows when not to use the HD map.
So basically it is constantly in the process of validating the prior scan to the actual existing in real time prior to making a decision. Seems awful lot of back and forth going on.
 
  • Funny
Reactions: AlanSubie4Life
Here are a few examples:
  • AV is driving. HD map matches the lane lines that the cameras, lidar and radar see. Cool. The car continues driving with increased confidence since both sources match.
  • The car can't see a stop sign because a storm knocked it down the day before but the car knows there is a stop sign there because of the HD map and slows down and stops.
  • There is a dense fog. The cameras and lidar have reduced range and can't see the lane lines very far ahead. But the HD map tells the car where the lanes should be, far ahead. The car can still drive despite the cameras and lidar not seeing as far.
  • AV is driving. HD map tells it that there is a sharp bend coming up soon that the sensors can't see. So the AV knows to slow down in anticipation of the bend in the road.
  • Traffic is congested. Cameras, lidar can't see all the lane lines because all the vehicles. HD map tells the car where the lanes are.
  • Car comes up to a construction zone. Cameras, lidar and radar detect the construction zone. The car overrides the HD map. The car navigates the construction zone based on the cameras, lidar and radar.
  • Cameras, lidar, radar detect a fallen tree. The car plots a safe path around the fallen tree using the cameras, lidar and radar.

So as you can see: the HD map can help the car in some instances but the car also knows when not to use the HD map.
All this is exactly what I said about the WRC rally team. Why would anyone stop at a stop sign that does not exist(anymore as of that day)? That is going to cause real trouble. Think hard about it.
 
All this is exactly what I said about the WRC rally team. Why would anyone stop at a stop sign that does not exist(anymore as of that day)? That is going to cause real trouble. Think hard about it.

No. Read what I wrote. The stop sign does exist, it just fell over. It is still a valid stop. The car still needs to stop for the sign. Imagine if the car just blew threw the intersection because it did not see the stop sign, that could cause a major accident.
 
Waymo is going down this road. Mark my words.


I see you are another Waymo doubter who thinks they will fail. Well, good luck with that. The Waymo doubters are like the TSLA shorts who are convinced Tesla will go bankrupt any day now. You keep waiting for Waymo to fail and they keep expanding more and more. If you are betting on Waymo failing, you will lose. Mark my words.