The Autoware Foundation
Multi LiDAR localization #2538
-
|
Does Autoware support localization using multiple lidar? Using Autoware, I am trying to localize a robot attached with two lidars (VLP-16), one in front and one in the back of the body. In the current Autoware structure, however, the localization nodes only use /sensing/lidar/top/* and not /sensing/lidar/concatenated/pointcloud. Can we just put this topic into localization nodes? Also, I would like to know if there is any reason why it doesn't use concatenated point cloud. Thank you in advance! |
Beta Was this translation helpful? Give feedback.
Replies: 0 comments 2 replies
-
|
Ideally, estimation by multiple Lidars is better because it increases the amount of information. In reality, however, the estimation may be unstable due to calibration errors and time synchronization discrepancies between Lidars. In many cases, a single Lidar will perform well, so I recommend using a single Lidar first. If you understand the problem with multiple Lidars, you can use Incidentally, if the vehicle is slow or if only the Lidar point cloud near the vehicle is used, this problem is not a major concern. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your answer! I am trying to apply Autoware on an indoor mobile robot that doesn't move so fast, so will also try using concatenated point cloud to see if it also works. |
Beta Was this translation helpful? Give feedback.
Ideally, estimation by multiple Lidars is better because it increases the amount of information.
In reality, however, the estimation may be unstable due to calibration errors and time synchronization discrepancies between Lidars.
In many cases, a single Lidar will perform well, so I recommend using a single Lidar first.
If you understand the problem with multiple Lidars, you can use
/sensing/lidar/concatenated/pointcloudas input.Incidentally, if the vehicle is slow or if only the Lidar point cloud near the vehicle is used, this problem is not a major concern.