Skip to content

Latest commit

 

History

History
71 lines (59 loc) · 2.46 KB

test_code.md

File metadata and controls

71 lines (59 loc) · 2.46 KB

Test code

✅ Dependencies

  • (All bbs3d dependencies)
  • PCL

📋 Test

1. Build

cd 3d_bbs/test/
mkdir build && cd build
  • CPU ver. & GPU ver.
cmake .. -DCMAKE_BUILD_TYPE=Release
make -j
  • CPU ver. only
cmake .. -DCMAKE_BUILD_TYPE=Release -DBUILD_CUDA=OFF
make -j

2. Download

Please download test data.

3. Config file setting

Config file format is 3d-bbs/test/config/test.yaml
Please edit the config file as below:

  1. Copy the target and source paths in the downloaded test_data to target_clouds and source_clouds items.
  2. Create the output folder where you want to save the output pcd and copy the path to output_folder in test.yaml.
  3. Test data work with default parameter values.

Alt text

4. Run

  • GPU
cd 3d_bbs/test/build/
./gpu_test <config_file_path>
  • CPU
cd 3d_bbs/test/build/
./cpu_test <config_file_path>

5. (Optional) Load voxelmap coordinates directly

You can save the voxelmaps coordinates and skip 3D-BBS voxel construction if you reuse the same parameters of min_level_res and max_level.

cd 3d_bbs/test/build/
./saver <target folder path> min_level_res max_level

Then, try again 4.Run. Voxelmaps coordinates in 'voxelmaps_coords' are automatically loaded.
Note: Test code preferentially uses the parameters in the generated text file 'voxel_params.txt'.

💻 Using your own 3D point cloud map and LiDAR scan data

Convert ros2 bag to pcd file

If you have your own ros2 bag data, you can convert LiDAR msgs to pcd file with package below so that point cloud aligns in the direction of gravitational acceleration using IMU msgs.
https://github.com/KOKIAOKI/ros2bag_to_pcd

Conditions for demonstrating 3D-BBS performance on your own data

  • Keep the robot with the sensor stationary.
    • Reason: The error in the direction of gravitational acceleration estimated by IMU increases while the robot is running.
    • Although the roll and pitch searches can be expanded, processing time will increase.
  • Use only at locations where the source point cloud is completely included in the target point cloud.
    • Reason: Another pose that encompasses all source point cloud is estimated when the source point cloud includes outside the map environment.
    • Please utilize the downsampling and point cloud cutting tools.