# **Soter Local Localization**
Local localization(pose tracking) by laserscan(lidar) and odometry, also needs map input.
****
## <font color=#FF0000>soter_local_localization_node</font>
****


### Published topics:
* **~scan** ( sensor_msgs/PointCloud2 )
Laserscan republished on map frame.
* **~scan_down** ( sensor_msgs/PointCloud2 )
Laserscan downsampled republished on map frame.
* **~potential_map** ( nav_msgs/OccupancyGrid )
Pub potential map if using gauss-newton.
* **~map_cloud** ( sensor_msgs/PointCloud2 )
Map cloud from occupancy grid map if using icp.
* **~submap_cloud** ( sensor_msgs/PointCloud2 )
Submap cloud cutted from whole map cloud if using icp.
### Subscribed Topics:
* **map_topic(from parameter setting)** ( nav_msgs/OccupancyGrid )
Map input for localization.
* **scan_topic(from parameter setting)** ( sensor_msgs/LaserScan )
Scan input if use_cloud_input is false.
* **scan_cloud_topic(from parameter setting)** ( sensor_msgs/PointCloud2 )
Scan input if use_cloud_input is true.
* **odom_topic(from parameter setting)** ( nav_msgs/Odometry )
Odom input if use_odom_input is true and odom_by_tf is false.
* **/initialpose** ( geometry_msgs::PoseWithCovarianceStamped )
For setting initial guess.

### Parameters
* **~map_topic:** (string, default:"/map")
Map topic to be subscribed.
* **~use_cloud_input:** (bool, default: false)
Whether using laser cloud input(2d).
* **~scan_topic:** (string, default:"/scan")
Scan topic to be subscribed if use_cloud_input is false.
* **~scan_cloud_topic:** (string, default:"/merged_cloud")
Scan cloud topic to be subscribed if use_cloud_input is true.
* #### Odometry input for scan matching initial guess
* **~use_odom_input:** (bool, default: true)
Whether using odometry as initial guess(no matter by tf tree or odom topic). If set to false, then just use laserscan to do localization.
* **~odom_by_tf:** (bool, default: true)
If set to true, then get initial guess from tf tree(If use_odom_input is true). If set to false, then get initial guess from odom topic.
* **~odom_topic:** (string, default:"odom")
Odom topic to be subscribed if using odom input and odom_by_tf is false.
* **~odom_frame_id:** (string, default:"odom")
For getting initial guess from tf tree if using odom input and odom_by_tf is true.
* **~robot_frame_id:** (string, default:"base_link")
For getting initial guess from tf tree if using odom input and odom_by_tf is true. And is also needed for transform laserscan from laser frame to robot frame
* **~published_frame_id:** (string, default:"odom")
Final localization result will be replaced by transform from map to published frame.
* **~transform_tolerance:** (double, default: 0.1)
Tf listener tolerance.
* **~transform_offset:** (double, default: 0.1)
For publishing tf in the future (for navigation).
* **~update_min_d:** (double, default: 0.02)
Do scan matching only when over this threshold(translation, if use odom).
* **~update_min_a:** (double, default: 0.02)
Do scan matching only when over this threshold(rotation, if use odom).
* **~halt_check_count:** (int, default: 10)
If over this count times not over update_min_a or update_min_a, then stop doing scan matching, just trust odom input
* #### Laserscan processor
* **~laser_min_range:** (double, default: -1)
-1 for using scan msg default setting.
* **~laser_max_range:** (double, default: -1)
-1 for using scan msg default setting.
* **~voxel_size:** (double, default: 0.2)
Voxel filter for downsample
* **~max_laser_points:** (int, default: 100)
Downsmaple after voxel filter.
* #### Scan matching
* **~scan_matching/scan_matching_method** (int, default: 1)
Scan matching method selection, 1 for gauss_newton, 2 for icp
* #### Gauss Newton
* **~scan_matching/gauss_newton/potential_map_resolution** (double, default: 0.05)
Potential map resolution.
* **~scan_matching/gauss_newton/likelihood_sigma** (double, default: 0.1)
For smoothing potential map and make it differentiable.
* **~scan_matching/gauss_newton/likelihood_max_dist** (double, default: 0.3)
For smoothing potential map and make it differentiable.
* **~scan_matching/gauss_newton/max_iteration** (int, default: 100)
Max iteration for gauss newton optimization.
* **~scan_matching/gauss_newton/step_size_scalar** (double, default: 0.1)
Slow down convergence speed.
* #### ICP
* **~scan_matching/icp/submap_size** (double, default: 10.0)
Cloud size of submap cutted from whole map.
* **~scan_matching/icp/submap_voxel_size** (double, default: 0.05)
For submap cloud downsample.
* **~scan_matching/icp/one_pixel_to_four_points** (bool, default: false)
Whether creating four points from one pixel for more accurate scan matching (higher compute error).
* **~scan_matching/icp/update_submap_dis** (double, default: 1.0)
Update submap cloud when moving over this distance.
* **~scan_matching/icp/max_iteration** (int, default: 100)
Max iteration of icp.
* **~scan_matching/icp/max_correspondence_distance** (double, default: 0.05)
Max correspondence distance after initial stage.
* **~scan_matching/icp/max_correspondence_distance_init** (double, default: 0.2)
Max correspondence distance for initial stage.
* **~scan_matching/icp/transformation_epsilon** (double, default: 1e-10)
Transformation epsilon.
* **~scan_matching/icp/euclidean_fitness_epsilon** (double, default: 0.1)
Euclidean fitness epsilon.

****
# **Demo Video**
ITRI lab test
[](https://www.youtube.com/watch?v=m0GKIkiNgxI&feature=youtu.be&ab_channel=%E6%9E%97%E6%94%BF%E7%A2%A9 "ITRI lab test")

ITRI dining room
[](https://www.youtube.com/watch?v=t4g4w9aQLjk&feature=youtu.be&ab_channel=%E6%9E%97%E6%94%BF%E7%A2%A9 "ITRI dining room")