Skip to content

PcdGeneration

Abstract

Pcd Generation is a tool for a vehicle based point cloud mapping in a simulation environment.
It is useful when you need a point cloud based on some location, but are not able to conduct physically mapping on the real place.

When Pcd Generation is conducted, Vehicle object (carrying LiDAR) is warped along all centerlines of lanelets in the imported OSM map. Point cloud map is generated to record points by LiDAR on Vehicle object on each centerlines of lanelets.

Overview

Traffic Simulation consists of the following flow:

graph TB
    subgraph A["PcdGeneration::Initialize()"]
        direction TB
        AA("Load LaneletMap")~~~AB("Create `Capture Pose Queue`")
    end

    AB-->BA

    subgraph B["PcdGeneration::OnUpdate()"]
        direction TB
        BA--"yes"-->BE("PcdGeneration::SavePcd()")
        BA{`Capture Pose Queue` is empty}--"no"-->BB("Dequeue `Capture Pose Queue`")-->BC("Move pose")-->BD("Capture by LiDAR")-->BA
    end

Configuration

PcdGeneration can be configured from PcdGenerator and RGLMappingAdapter components.

The configurable elements are listed in the following table:

PcdGenerator

Parameter Description
Osm Data Container Imported OSM file.
Vehicle Transform Game object containing sensors to capture pointcloud.
Rgl Mapping Adapter Reference of LiDAR object which is attached RGLMappingAdapter.
Output Pcd File Path Result PCD file name.
On captured, it will be saved in Assets/[Output Pcd File Path]
Capture Location Interval Distance in meters between consecutive warps along the centerline of a lanelet.
World Origin ROS World origin in ROS coordinate systems, will be added to every point coordinates.

RGLMappingAdapter

Parameter Description
Enable Downsampling Enable/disable point cloud data downsampling.
Leaf Size Resolution of point cloud data downsampling.

Execution and Parameters

Execution

If you play prepared scene, PcdGenerator will automatically start mapping.
PCD file will be written when you stop scene or all lanelets in the route are captured.

If the Vehicle stops moving and you see the following message in the bottom left corner, you can safely stop the scene.

pcd save success

The Point cloud (*.pcd) file is saved to the location you specified in the Output Pcd File Path of PcdGenerator.

Parameters

Following parameters are useful to point cloud map of quality and file size you want to.

Leaf Size

Downsampling aims to reduce PCD size which for large point clouds may achieve gigabytes in exchange for map details.
It is essential to find the acceptable balance between the file size and details level.

Leaf Size can be configured by RGLMappingAdapter component.

A small Leaf Size results in a more detailed PCD, while a large Leaf Size could result in excessive filtering such that objects like buildings are not recorded in the PCD.

Leaf Size = 1.0 Leaf Size = 10.0 Leaf Size = 100.0
Leaf Size 1 Leaf Size 10 Leaf Size 100 \

Capture Location Interval

If the Capture Location Interval is too small, it could result in a sparse PCD where some region of the map is captured well but the other regions aren't captured at all.

Capture Location Interval can be configured by PcdGeneration component.

Capture Location Interval = 6 Capture Location Interval = 20 Capture Location Interval = 100
Interval 6 Interval 20 Interval 100

Instruction

To use Pcd Generation, please follow the steps below.

For the preparation, the following must be prepared:

  • 3D map (.fbx)
  • lanelet map (.osm)
  • LiDAR sensor (.fbx)

Info

AWSIM includes AutowareSimulationDemo scene.
Please refer to:
* Assets/Awsim/Scenes/PcdGenerationDemo.unity

1. Add a Vehicle object

Add Vehicle object to carry LiDAR and capture points.
In addition, visual elements of Vehicle if needed.

Please create Vehicle object as the following:

  1. Create empty GameObject (should be Vehicle)
  2. (optional) Create empty GameObject (should be Vehicle/Geometry)
    1. If needed, visual elements of Vehicle added here
    2. Visual element can even be a simple Cube object as the sample image

2. Add a Camera

Add Unity Camera to visualize recording process.

Please create Unity Camera as the following:

  1. Create Camera object (should be Vehicle/Follow Camera)
  2. Attach FollowCamera component to Follow Camera
  3. Fill in Target field with Vehicle
  4. Modify Transform
    1. Using Cameras Unity view allows you to check the camera view

Add LiDAR object and configure components to record points.

LiDAR object

Please create and configure LiDAR object as the following:

  1. Create LiDAR object (should be Vehicle/LiDAR)
  2. Set Transform.Position.Y 2.5
  3. Attach PointCloudVisualization component to LiDAR
  4. Attach LidarSensor component to LiDAR
    1. Select Model Preset (VelodyneVLP16 and VelodyneVLS128 is recommended)
      1. VelodyneVLS128 can create more detail maps than VelodyneVLP16
      2. VelodyneVLP16 can work lighter than VelodyneVLS128
    2. Set Apply Distance Gaussian Noise and Apply Angular Gaussian Noise False
  5. Attach RGLMappingAdapter component to LiDAR

RGLSceneManager

Please create SceneManager object as the following:

  1. Create empty GameObject (should be RGLSceneManager)
  2. Attach SceneManager component to RGLSceneManager

4. Setup PcdGenerator

Add PcdGenerator component to manage above objects and create point cloud map.

  1. Create empty GameObject (should be PcdGenerator)
  2. Attach PcdGenerator component to PcdGenerator
  3. Fill in Osm Data Container field with .osm file
    1. If there is not .osm file in project, move .osm file to Assets/Awsim/Externals directory using file expoler
  4. Fill in Vehicle Transform field with Vehicle
  5. Fill in Rgl Mapping Adapter field with LiDAR
  6. (optional) Fill in World Origin ROS field if your map has Mgrs Position component

5. Call methods of PcdGenerator and FollowCamera

Some methods of PcdGenerator and FollowCamera should be called from callback of MonoBehaviour to enable Pcd Generation.

Please implement as the following:

  1. Create or open class which is inherit MonoBehaviour
  2. Make fields of PcdGenerator and FollowCamera
  3. Add description of calling method of PcdGenerator and FollowCamera

The method should be called are listed in the following table:

Pcd Generation

Method Description
Initialize() Should be called Start() callback.
OnUpdate() Should be called Update() callback.

FollowCamera

Method Description
Initialize() Should be called Start() callback.
OnUpdate() Should be called Update() callback.

Info

AWSIM includes AutowareSimulationDemo scene.
Please refer to:
* Assets/Awsim/Scenes/PcdGenerationDemo/PcdGenerationDemo.cs
* Assets/Awsim/Scenes/PcdGenerationDemo.unity scene

Modify and Verify PCD files

Modify PCD

Generated PCD files should be downsample because they are typically too large.
In addition, they should be converted to ASCII format because Autoware accepts only this format. (Pcd Generation returns PCD in binary format)

The tool (DownsampleLargePCD) can realize downsampling and converting PCD files. Please clone and build DownsampleLargePCD tool following Github repository.

Please use DownsampleLargePCD tool as the following:

  1. Change the working directory to the location with DownsampleLargePCD tool.
  2. Use DownsampleLargePCD tool to downsample and save PCD in ASCII format. 
    ./DownsampleLargePCD -in <PATH_TO_INPUT_PCD> -out <PATH_TO_OUTPUT_PCD> -leaf 0.2,0.2,0.2
    • You can also save PCD in binary format by adding -binary 1 option.

Note

If you don't want to downsample but want to convert PCD file to ASCII, you should use pcl_convert_pcd_ascii_binary tool.
This tool is available in the pcl-tools package and can be installed on Ubuntu with the following command: 

sudo apt install pcl-tools
To convert your PCD use command: 
pcl_convert_pcd_ascii_binary <PATH_TO_INPUT_PCD> <PATH_TO_OUTPUT_PCD> 0

Verify PCD

To verify your PCD you can launch the Autoware*(https://github.com/autowarefoundation/autoware).

Please verify PCD as the following:

  1. Copy your PCD from the AWSIM project directory to the Autoware map directory 

    cp <PATH_TO_PCD_FILE> <PATH_TO_AUTOWARE_MAP>/

  2. Source the ROS and Autoware 

    source /opt/ros/humble/setup.bash
source <PATH_TO_AUTOWARE>/install/setup.bash

  3. Launch the planning simulation with the map directory path (map_path) and PCD file (pointcloud_map_file) specified 

    ros2 launch autoware_launch planning_simulator.launch.xml \
vehicle_model:=sample_vehicle \
sensor_model:=sample_sensor_kit \
map_path:=<ABSOLUTE_PATH_TO_AUTOWARE_MAP> \
pointcloud_map_file:=<PCD_FILE_NAME>

    PCD file location

    The PCD file needs to be located in the Autoware map directory and as a pointcloud_map_file parameter you only supply the file name, not the path.

    Absolute path

    When launching Autoware never use ~/ to specify the home directory. Either write the full absolute path or use $HOME environmental variable.

  4. Wait for the Autoware to finish loading and inspect the PCD visually given the effect of Leaf Size and effect of Capture Location Interval.