Skip to content

Commit

Permalink
refactor(ndt_scan_matcher): rework parameters (autowarefoundation#6229)
Browse files Browse the repository at this point in the history
* refactor(ndt_scan_matcher): rework parameters

Signed-off-by: Yamato Ando <yamato.ando@tier4.jp>

* delete tmp file

Signed-off-by: Yamato Ando <yamato.ando@tier4.jp>

* style(pre-commit): autofix

---------

Signed-off-by: Yamato Ando <yamato.ando@tier4.jp>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
  • Loading branch information
2 people authored and anhnv3991 committed Feb 13, 2024
1 parent fb14017 commit 693fb8d
Show file tree
Hide file tree
Showing 12 changed files with 390 additions and 40 deletions.
67 changes: 27 additions & 40 deletions localization/ndt_scan_matcher/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -56,28 +56,29 @@ One optional function is regularization. Please see the regularization chapter i

### Core Parameters

| Name | Type | Description |
| --------------------------------------------------------- | ---------------------- | -------------------------------------------------------------------------------------------------- |
| `base_frame` | string | Vehicle reference frame |
| `ndt_base_frame` | string | NDT reference frame |
| `map_frame` | string | map frame |
| `input_sensor_points_queue_size` | int | Subscriber queue size |
| `trans_epsilon` | double | The max difference between two consecutive transformations to consider convergence |
| `step_size` | double | The newton line search maximum step length |
| `resolution` | double | The ND voxel grid resolution [m] |
| `max_iterations` | int | The number of iterations required to calculate alignment |
| `converged_param_type` | int | The type of indicators for scan matching score (0: TP, 1: NVTL) |
| `converged_param_transform_probability` | double | TP threshold for deciding whether to trust the estimation result (when converged_param_type = 0) |
| `converged_param_nearest_voxel_transformation_likelihood` | double | NVTL threshold for deciding whether to trust the estimation result (when converged_param_type = 1) |
| `initial_estimate_particles_num` | int | The number of particles to estimate initial pose |
| `n_startup_trials` | int | The number of initial random trials in the TPE (Tree-Structured Parzen Estimator). |
| `lidar_topic_timeout_sec` | double | Tolerance of timestamp difference between current time and sensor pointcloud |
| `initial_pose_timeout_sec` | int | Tolerance of timestamp difference between initial_pose and sensor pointcloud. [sec] |
| `initial_pose_distance_tolerance_m` | double | Tolerance of distance difference between two initial poses used for linear interpolation. [m] |
| `num_threads` | int | Number of threads used for parallel computing |
| `output_pose_covariance` | std::array<double, 36> | The covariance of output pose |

(TP: Transform Probability, NVTL: Nearest Voxel Transform Probability)
#### Frame

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/frame.json") }}

#### Ndt

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/ndt.json") }}

#### Initial Pose Estimation

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/initial_pose_estimation.json") }}

#### Validation

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/validation.json") }}

#### Score Estimation

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/score_estimation.json") }}

#### Covariance

{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/covariance.json") }}

## Regularization

Expand Down Expand Up @@ -153,10 +154,7 @@ This is because if the base position is far off from the true value, NDT scan ma

### Parameters

| Name | Type | Description |
| ----------------------------- | ------ | ---------------------------------------------------------------------- |
| `regularization_enabled` | bool | Flag to add regularization term to NDT optimization (FALSE by default) |
| `regularization_scale_factor` | double | Coefficient of the regularization term. |
{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/ndt_regularization.json") }}

Regularization is disabled by default because GNSS is not always accurate enough to serve the appropriate base position in any scenes.

Expand Down Expand Up @@ -206,11 +204,7 @@ Using the feature, `ndt_scan_matcher` can theoretically handle any large size ma

### Parameters

| Name | Type | Description |
| ------------------------------------- | ------ | ------------------------------------------------------------ |
| `dynamic_map_loading_update_distance` | double | Distance traveled to load new map(s) |
| `dynamic_map_loading_map_radius` | double | Map loading radius for every update |
| `lidar_radius` | double | LiDAR radius used for localization (only used for diagnosis) |
{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/dynamic_map_loading.json") }}

### Notes for dynamic map loading

Expand All @@ -237,10 +231,7 @@ This is a function that uses no ground LiDAR scan to estimate the scan matching

### Parameters

| Name | Type | Description |
| ------------------------------------- | ------ | ----------------------------------------------------------------------------------- |
| `estimate_scores_by_no_ground_points` | bool | Flag for using scan matching score based on no ground LiDAR scan (FALSE by default) |
| `z_margin_for_ground_removal` | double | Z-value margin for removal ground points |
{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/score_estimation_no_ground_points.json") }}

## 2D real-time covariance estimation

Expand All @@ -259,8 +250,4 @@ Note that this function may spoil healthy system behavior if it consumes much ca
initial_pose_offset_model is rotated around (x,y) = (0,0) in the direction of the first principal component of the Hessian matrix.
initial_pose_offset_model_x & initial_pose_offset_model_y must have the same number of elements.

| Name | Type | Description |
| ----------------------------- | ------------------- | ----------------------------------------------------------------- |
| `use_covariance_estimation` | bool | Flag for using real-time covariance estimation (FALSE by default) |
| `initial_pose_offset_model_x` | std::vector<double> | X-axis offset [m] |
| `initial_pose_offset_model_y` | std::vector<double> | Y-axis offset [m] |
{{ json_to_markdown("localization/ndt_scan_matcher/schema/sub/covariance_covariance_estimation.json") }}
44 changes: 44 additions & 0 deletions localization/ndt_scan_matcher/schema/ndt_scan_matcher.schema.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,44 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"type": "object",
"properties": {
"/**": {
"type": "object",
"properties": {
"ros__parameters": {
"type": "object",
"properties": {
"frame": { "$ref": "sub/frame.json#/definitions/frame" },
"ndt": { "$ref": "sub/ndt.json#/definitions/ndt" },
"regularization": { "$ref": "ndt_regularization.json#/definitions/ndt/regularization" },
"initial_pose_estimation": {
"$ref": "sub/initial_pose_estimation.json#/definitions/initial_pose_estimation"
},
"validation": { "$ref": "sub/validation.json#/definitions/validation" },
"score_estimation": {
"$ref": "sub/score_estimation.json#/definitions/score_estimation"
},
"covariance": { "$ref": "sub/covariance.json#/definitions/covariance" },
"dynamic_map_loading": {
"$ref": "sub/dynamic_map_loading.json#/definitions/dynamic_map_loading"
}
},
"required": [
"frame",
"ndt",
"initial_pose_estimation",
"validation",
"score_estimation",
"covariance",
"dynamic_map_loading"
],
"additionalProperties": false
}
},
"required": ["ros__parameters"],
"additionalProperties": false
}
},
"required": ["/**"],
"additionalProperties": false
}
25 changes: 25 additions & 0 deletions localization/ndt_scan_matcher/schema/sub/covariance.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"covariance": {
"type": "object",
"properties": {
"output_pose_covariance": {
"type": "array",
"description": "The covariance of output pose. Note that this covariance matrix is empirically derived.",
"default": [
0.0225, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0225, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0225, 0.0,
0.0, 0.0, 0.0, 0.0, 0.0, 0.000625, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.000625, 0.0, 0.0,
0.0, 0.0, 0.0, 0.0, 0.000625
]
},
"covariance_estimation": {
"$ref": "covariance_covariance_estimation.json#/definitions/covariance_estimation"
}
},
"required": ["output_pose_covariance", "covariance_estimation"],
"additionalProperties": false
}
}
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"covariance_estimation": {
"type": "object",
"properties": {
"enable": {
"type": "boolean",
"description": "2D Real-time covariance estimation with multiple searches (output_pose_covariance is the minimum value).",
"default": false
},
"initial_pose_offset_model_x": {
"type": "array",
"description": "Offset arrangement in covariance estimation [m]. initial_pose_offset_model_x & initial_pose_offset_model_y must have the same number of elements.",
"default": [0.0, 0.0, 0.5, -0.5, 1.0, -1.0]
},
"initial_pose_offset_model_y": {
"type": "array",
"description": "Offset arrangement in covariance estimation [m]. initial_pose_offset_model_x & initial_pose_offset_model_y must have the same number of elements.",
"default": [0.5, -0.5, 0.0, 0.0, 0.0, 0.0]
}
},
"required": ["enable", "initial_pose_offset_model_x", "initial_pose_offset_model_y"],
"additionalProperties": false
}
}
}
31 changes: 31 additions & 0 deletions localization/ndt_scan_matcher/schema/sub/dynamic_map_loading.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,31 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"dynamic_map_loading": {
"type": "object",
"properties": {
"update_distance": {
"type": "number",
"description": "Dynamic map loading distance.",
"default": 20.0,
"minimum": 0.0
},
"map_radius": {
"type": "number",
"description": "Dynamic map loading loading radius.",
"default": 150.0,
"minimum": 0.0
},
"lidar_radius": {
"type": "number",
"description": "Radius of input LiDAR range (used for diagnostics of dynamic map loading).",
"default": 100.0,
"minimum": 0.0
}
},
"required": ["update_distance", "map_radius", "lidar_radius"],
"additionalProperties": false
}
}
}
28 changes: 28 additions & 0 deletions localization/ndt_scan_matcher/schema/sub/frame.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"frame": {
"type": "object",
"properties": {
"base_frame": {
"type": "string",
"description": "Vehicle reference frame.",
"default": "base_link"
},
"ndt_base_frame": {
"type": "string",
"description": "NDT reference frame.",
"default": "ndt_base_link"
},
"map_frame": {
"type": "string",
"description": "Map frame.",
"default": "map"
}
},
"required": ["base_frame", "ndt_base_frame", "map_frame"],
"additionalProperties": false
}
}
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"initial_pose_estimation": {
"type": "object",
"properties": {
"particles_num": {
"type": "number",
"description": "The number of particles to estimate initial pose.",
"default": 200,
"minimum": 1
},
"n_startup_trials": {
"type": "number",
"description": "The number of initial random trials in the TPE (Tree-Structured Parzen Estimator). This value should be equal to or less than 'initial_estimate_particles_num' and more than 0. If it is equal to 'initial_estimate_particles_num', the search will be the same as a full random search.",
"default": 20,
"minimum": 1
}
},
"required": ["particles_num", "n_startup_trials"],
"additionalProperties": false
}
}
}
53 changes: 53 additions & 0 deletions localization/ndt_scan_matcher/schema/sub/ndt.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,53 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"ndt": {
"type": "object",
"properties": {
"trans_epsilon": {
"type": "number",
"description": "The maximum difference between two consecutive transformations in order to consider convergence.",
"default": 0.01,
"minimum": 0.0
},
"step_size": {
"type": "number",
"description": "The newton line search maximum step length.",
"default": 0.1,
"minimum": 0.0
},
"resolution": {
"type": "number",
"description": "The ND voxel grid resolution.",
"default": 2.0,
"minimum": 0.0
},
"max_iterations": {
"type": "number",
"description": "The number of iterations required to calculate alignment.",
"default": 30,
"minimum": 1
},
"num_threads": {
"type": "number",
"description": "Number of threads used for parallel computing.",
"default": 4,
"minimum": 1
},
"regularization": {
"$ref": "ndt_regularization.json#/definitions/regularization"
}
},
"required": [
"trans_epsilon",
"step_size",
"resolution",
"max_iterations",
"num_threads",
"regularization"
],
"additionalProperties": false
}
}
}
24 changes: 24 additions & 0 deletions localization/ndt_scan_matcher/schema/sub/ndt_regularization.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Parameters for Ndt Scan Matcher Node",
"definitions": {
"regularization": {
"type": "object",
"properties": {
"enable": {
"type": "boolean",
"description": "Regularization switch.",
"default": false
},
"scale_factor": {
"type": "number",
"description": "Regularization scale factor.",
"default": 0.01,
"minimum": 0.0
}
},
"required": ["enable", "scale_factor"],
"additionalProperties": false
}
}
}
Loading

0 comments on commit 693fb8d

Please sign in to comment.