Skip to content

Conversation

@technolojin
Copy link
Contributor

Description

Related links

Parent Issue:

  • Link

How was this PR tested?

Notes for reviewers

None.

Interface changes

None.

Effects on system behavior

None.

@technolojin technolojin self-assigned this Nov 19, 2025
@github-actions github-actions bot added component:perception Advanced sensor data processing and environment understanding. (auto-assigned) component:launch Launch files, scripts and initialization tools. (auto-assigned) labels Nov 19, 2025
@github-actions
Copy link

github-actions bot commented Nov 19, 2025

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

technolojin and others added 7 commits November 19, 2025 11:02
…onality

- Introduced `detectVoxelsDebug` method to visualize voxel generation without running inference.
- Added `generateVoxelBoxes3D_launch` method for launching the CUDA kernel to create 3D boxes from voxel coordinates.
- Updated `postProcess` to include voxel visualization logging and checks for the number of pillars.
…ion filtering

- Added `is_valid_voxel` struct to filter out invalid boxes based on score.
- Introduced visualization bounds for filtering boxes outside specified ranges.
- Updated `generateVoxelBoxes3D_kernel` to use a minimal height for ground-level visualization.
- Modified `generateVoxelBoxes3D_launch` to copy only valid boxes to the host, improving efficiency.
…ization methods

- Deleted `detectVoxelsDebug` method and associated CUDA kernel for generating voxel boxes, as they are no longer needed.
- Removed `generateVoxelBoxes3D_launch` method and its related kernel to streamline the codebase.
- Updated visualization bounds in `pointCloudCallback` for improved range handling.
- Adjusted occupancy mapping logic to reflect changes in visualization requirements.
…on support

- Introduced `getVoxelGridData` method to retrieve voxel heights along with coordinates and point counts.
- Implemented `computeVoxelHeights_launch` CUDA kernel to calculate the height of each voxel based on its points.
- Updated `pointCloudCallback` to support visualization mode selection between occupancy and height.
- Enhanced data handling for voxel heights in the visualization process, improving the representation of voxel data.
- Introduced `getVoxelGridData` method to retrieve voxel mean Z values along with coordinates, point counts, and voxel heights.
- Implemented `extractFeatureMeanZ_launch` CUDA kernel to compute mean Z from encoded features.
- Updated `pointCloudCallback` to include a new visualization mode for mean Z, enhancing data representation.
- Enhanced memory management for voxel mean Z in the processing pipeline, improving overall efficiency.
@github-actions github-actions bot removed the component:launch Launch files, scripts and initialization tools. (auto-assigned) label Nov 19, 2025
technolojin and others added 2 commits November 19, 2025 13:16
- Implemented transformation of the occupancy grid from the ego frame to the map frame using TF2.
- Added error handling to fallback to the ego frame if the transformation fails.
- Updated grid message header and origin position based on the transformed coordinates, enhancing spatial accuracy in occupancy mapping.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

component:perception Advanced sensor data processing and environment understanding. (auto-assigned)

Projects

Status: To Triage

Development

Successfully merging this pull request may close these issues.

1 participant