Eagleye (GNSS+IMU) localization algorithm results in offset #11431
Description
Checklist
- I've read the contribution guidelines.
- I've searched other issues and no duplicate issues were found.
- I'm convinced that this is not my fault but a bug.
Description
I am facing an issue with the Eagleye (GNSS + IMU) localization algorithm in Autoware Universe. When running Eagleye localization, there is always a noticeable offset in the vehicle position, the LiDAR point cloud does not align properly with the map (as shown in the attached video). The transforms between all sensors are correct, and even after manually adjusting them, the localization offset remained unchanged. The video might seem a bit laggy, with occasional sudden jumps in the vehicle’s position because of a video recording issue, but the localization itself remained smooth.
Expected behavior
After the calibration step, the eagleye algorithm should provide the pose and twist for localization and the LiDAR point cloud should align with the map during localization.
Actual behavior
The LiDAR point cloud shows an offset relative to the map. Adjusting sensor extrinsics does not resolve the issue.
Steps to reproduce
Repository for reproducing the issue
This repository includes the README with detailed reproduction steps and the required files.
Google Drive for ROS2 Bag, map and video data
The Google Drive link contains a .zip file. After extracting it, you’ll find the ROS2 bag, map data, and video.
Versions
Autoware: I wasn’t able to find an official release corresponding to my current setup, but the code is at commit 5ee2907 (latest commit as of 20 May 2024). Hopefully, this helps in identifying the version used.
ROS2: Humble
OS: Ubantu 22.04
Possible causes
No response
Additional context
No response