You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi all, I'm trying to make an "ADAS+" system by strapping my car with senors and using Autoware(the idea mainly coming from seeing Comma and wondering if we can do better with more sensors), however, I notice that the current way Autoware works is by using a point cloud map. Which wouldn't really work for this purpose given that you can't really pre-map places you haven't gone to yet. So I'm exploring the alternatives.
Here are just some random thoughts of mine(please excuse the ignorance, I really don't know what I am doing here)
I have found YabLoc from Autoware.Universe, which seems like a good choice, but I wonder if there's a better way(more accurate?) way to do this.
It also requires a vector map, which I am not sure if I will be able to provide? so it won't work if I can't directly use data from a map source like OSM.
Given the namesake of SLAM, would it be perhaps be possible to be used for localization instead?
Localization seems to be a smaller issue compared to other systems that may use the map? Given there are so many mechanisms for localization, if visual SLAM works, wouldn't LiDAR SLAM work the same/better?
I have also came across this discussion, which seems to be very close to what I'm going for, but it doesn't seem to be ready yet.
It seems like if I were to have a Vector map, the path is pretty open. Given that the design page indicates that it's used for localization and perception, both of which I feel like could be replaced one way or another. But I am not sure if I can even get that info if it can't be derived from map data(can I? Please advise).
But overall, it seems kind of feasible to crack together something that'll make it get moving, and maybe if we strap enough sensors onto it, it can be semi decent? I feel like I'm overlooking things, but I don't understand enough of the system to know what the showstopper(if any) here is.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Code of Conduct
Contents
Hi all, I'm trying to make an "ADAS+" system by strapping my car with senors and using Autoware(the idea mainly coming from seeing Comma and wondering if we can do better with more sensors), however, I notice that the current way Autoware works is by using a point cloud map. Which wouldn't really work for this purpose given that you can't really pre-map places you haven't gone to yet. So I'm exploring the alternatives.
Here are just some random thoughts of mine(please excuse the ignorance, I really don't know what I am doing here)
YabLoc
from Autoware.Universe, which seems like a good choice, but I wonder if there's a better way(more accurate?) way to do this.It seems like if I were to have a Vector map, the path is pretty open. Given that the design page indicates that it's used for localization and perception, both of which I feel like could be replaced one way or another. But I am not sure if I can even get that info if it can't be derived from map data(can I? Please advise).
But overall, it seems kind of feasible to crack together something that'll make it get moving, and maybe if we strap enough sensors onto it, it can be semi decent? I feel like I'm overlooking things, but I don't understand enough of the system to know what the showstopper(if any) here is.
Please go easy on me.
Beta Was this translation helpful? Give feedback.
All reactions