yabloc_pose_initializer#
This package contains a node related to initial pose estimation.
This package requires the pre-trained semantic segmentation model for runtime. This model is usually downloaded by the ansible artifacts role during the installation. It is also possible to download it manually. Even if the model is not downloaded, initialization will still complete, but the accuracy may be compromised.
The model is hosted on Hugging Face. To download it manually with the hf CLI:
hf download AutowareFoundation/yabloc_pose_initializer --revision v1.0 --local-dir ~/autoware_data/ml_models/yabloc_pose_initializer
Note#
This package makes use of external code. The trained files are provided by apollo. The trained files are automatically downloaded during env preparation.
Original model URL
Open Model Zoo is licensed under Apache License Version 2.0.
Converted model URL
https://github.com/PINTO0309/PINTO_model_zoo/tree/main/136_road-segmentation-adas-0001
model conversion scripts are released under the MIT license
Special thanks#
camera_pose_initializer#
Purpose#
- This node estimates the initial position using the camera at the request of ADAPI.
Input#
| Name | Type | Description |
|---|---|---|
input/camera_info |
sensor_msgs::msg::CameraInfo |
undistorted camera info |
input/image_raw |
sensor_msgs::msg::Image |
undistorted camera image |
input/vector_map |
autoware_map_msgs::msg::LaneletMapBin |
vector map |
Output#
| Name | Type | Description |
|---|---|---|
debug/init_candidates |
visualization_msgs::msg::MarkerArray |
initial pose candidates |
Parameters#
| Name | Type | Description | Default | Range |
|---|---|---|---|---|
| angle_resolution | float | how many divisions of 1 sigma angle range | 30 | N/A |
Services#
| Name | Type | Description |
|---|---|---|
yabloc_align_srv |
autoware_internal_localization_msgs::srv::PoseWithCovarianceStamped |
initial pose estimation request |