This project provides wall localization functionality based on laser scanners, suitable for scenarios where a cleaning robot performs wall-following cleaning tasks.
mkdir -p catkin_ws/src && cd catkin_ws/src
git clone https://github.com/SheldonFung98/WallStalker-Lidar.git
cd ..
catkin_make
source devel/setup.sh
roslaunch laser_stickwall detect_wall.launch
- Region of Interest (ROI) Extraction: The laser scanner's ROI is defined as 2 meters in front, 0.5 meters behind, and 1 meter on each side of the robot's laser origin. This can be configured in
laser_stickwall/launch/detect_wall.launch. - Rough Wall Matching: Linear regression is performed on all points within the ROI to approximate the wall. The quality of the fit is determined based on the point-to-line distance error.
- Obstacle Removal: Starting from the foremost point among the roughly matched wall points, obstacles (points with a distance too far from the fitted line) are removed to obtain valid wall points.
- Fine Wall Matching: Another linear regression is applied to the valid wall points to obtain a more accurate representation of the wall.
- Navigation Point Generation: Uniformly spaced navigation points are generated along the wall line.
- Visualization information can be obtained during debugging by subscribing to the topic
/elevator/markerArray. - Visualized images can be obtained during debugging by subscribing to the topic
/elevator/image.
- At the starting point of the wall-following task, the robot should be close to the wall (within 1 meter).
- The wall from the starting point to the endpoint should be at least 2 meters long.
| Topic Name | Type | Description |
|---|---|---|
| /scan | sensor_msgs/LaserScan | Laser scanner data |
| Topic Name | Type | Description |
|---|---|---|
| /wall_nav/wall_path | nav_msgs::Path | Wall-following navigation information |
| /wall_nav/image | sensor_msgs::Image | Image for debugging |