Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |
System Dependencies
Dependant Packages
Launch files
Messages
Services
Plugins
Recent questions tagged autoware_cuda_utils at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 0.48.0 |
| License | Apache License 2.0 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | |
| Checkout URI | https://github.com/autowarefoundation/autoware_universe.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-12-03 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | planner ros calibration self-driving-car autonomous-driving autonomous-vehicles ros2 3d-map autoware |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Manato Hirabayashi
- Kenzo Lobos-Tsunekawa
- Amadeusz Szymko
Authors
- Daisuke Nishimatsu
autoware_cuda_utils
Purpose
This package contains a library of common functions related to CUDA.
Changelog for package autoware_cuda_utils
0.48.0 (2025-11-18)
-
Merge remote-tracking branch 'origin/main' into humble
-
feat(autoware_cuda_pointcloud_preprocessor): cuda/polar voxel filter (#11122)
- feat(cuda_utils): support device memory allocation from memory pool
* feat(cuda_pointcloud_preprocessor): support polar_voxel_outlier_filter WIP: use cuda::std::optional. compile passed wip: version 1 wip: update wip: update launcher
- feat(cuda_pointcloud_preprocessor): add a flag to enable/disable ring outlier filter in cuda_pointcloud_preprocessor
- chore: clean up the code
- docs: update documents briefly
- style(pre-commit): autofix
- style(pre-commit): autofix
- feat(cuda_polar_voxel_outlier_filter): add xyzirc format support
- fix(cuda_polar_voxel_outlier_filter): move sync point to avoid unexpected memory release during async copy
- chore(cuda_polar_voxel_outlier_filter): update parameters
- add SI unit postfix
- deprecate [secondary_return_type]{.title-ref}
- and think points with non-primary return value as points with secondary return
- refactor(cuda_polar_voxel_outlier_filter): explicity specify index integer type
- refactor(cuda_polar_voxel_outlier_filter): snake case for functions
* refactor(cuda_polar_voxel_outlier_filter): std::optional for visibility and filter ratio And update related task functions for diagnostics
- fix(cuda_polar_voxel_outlier_filter): register parameters_callback
- refactor(cuda_polar_voxel_outlier_filter): remove log spam and unneccesary comments
- refactor(cuda_polar_voxel_outlier_filter): rename [valid_points_mask]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): make noise pointcloud publishing optional Because all parameters are now compatible with [autoware_pointcloud_preprocessor::polar_voxel_outlier_filter]{.title-ref}, this commit also removes the parameter files named [cuda_polar_voxel_outlier_filter.param.yaml]{.title-ref} to avoid duplicated file copying.
- refactor(cuda_polar_voxel_outlier_filter): simplify by enforcing use of XYZIRC or XYZIRCAEDT
- feat(cuda_polar_voxel_outlier_filter): limit range in visibility calculation
- refactor(cuda_polar_voxel_outlier_filter): use array of int for return_type instead of int64
* refactor(cuda_polar_voxel_outlier_filter): update parameter callback to align with the CPU implementation And small clean up the codes
* fix(cuda_polar_voxel_outlier_filter): return when invalid index fix the error revealed by [compute-sanitizer --tool memcheck]{.title-ref}
* feat(cuda_polar_voxel_outlier_filter): add visibility estimation parameters And update visibility calculation to align with the CPU implementation
- feat(cuda_polar_voxel_outlier_filter): add option to not publish a filtered pointcloud (only estimate visibility)
- fix(cuda_polar_voxel_outlier_filter): ensure zero started positive values for indices
- feat(cuda_polar_voxel_outlier_filter): add input validation. align diag format to the CPU implementation
- perf(cuda_polar_voxel_outlier_filter): skip output generation if visualization_estimation_only==true
- refactor(cuda_polar_voxel_outlier_filter): clean up the code
- feat(cuda_polar_voxel_outlier_filter): add intensity parameter for secondary returns
- fix(cuda_polar_voxel_outlier_filter): update param name to align CPU impl.
- feat(cuda_polar_voxel_outlier_filter): update codes to align CPU impl.
- fix(cuda_polar_voxel_outlier_filter): correct unintended comparison
- fix(cuda_polar_voxel_outlier_filter): correct meaningless cast
- refactor(cuda_polar_voxel_outlier_filter): unify common calculation
* chore(cuda_polar_voxel_outlier_filter): use auto for CUDA thread index As CUDA grid/block/thread indices, such as threadIdx are defined using unsigned int, using size_t is overkill
- docs(cuda_polar_voxel_outlier_filter): add description for numerical discrepancies
- fix(cuda_polar_voxel_outlier_filter): guard processing if input
File truncated at 100 lines see the full file
Package Dependencies
| Deps | Name |
|---|---|
| ament_cmake_auto | |
| autoware_cmake | |
| ament_cmake_gtest | |
| ament_lint_auto | |
| autoware_lint_common | |
| autoware_cuda_dependency_meta |