cuda_voxel_grid_downsample_filter
Purpose
This node is a CUDA accelerated version of the FasterVoxelGridDownsampleFilter available in autoware_cuda_pointcloud_preprocessor.
Inner-workings / Algorithms
This node reimplements of the function of autoware::pointcloud_preprocessor::FasterVoxelGridDownsampleFilter, which calculates centroids as representative points in each voxels.
| Name |
Type |
Description |
~/input/pointcloud |
sensor_msgs::msg::PointCloud2 |
Input pointcloud's topic. |
~/input/pointcloud/cuda |
negotiated_interfaces/msg/NegotiatedTopicsInfo |
Input pointcloud's type negotiation topic |
Output
| Name |
Type |
Description |
~/output/pointcloud |
sensor_msgs::msg::PointCloud2 |
Processed pointcloud's topic (in PointXYZIRC fashion) |
~/output/pointcloud/cuda |
negotiated_interfaces/msg/NegotiatedTopicsInfo |
Processed pointcloud's negotiation topic |
Parameters
Core Parameters
| Name |
Type |
Description |
Default |
Range |
| voxel_size_x |
float |
the voxel size along x-axis [m] |
0.3 |
≥0.0 |
| voxel_size_y |
float |
the voxel size along y-axis [m] |
0.3 |
≥0.0 |
| voxel_size_z |
float |
the voxel size along z-axis [m] |
0.1 |
≥0.0 |
| max_mem_pool_size_in_byte |
float |
maximum size of GPU memory pool |
1e+09 |
≥0 |
Assumptions / Known limits
- This node expects that the input pointcloud is compatible with
autoware::point_types::PointXYZI (ref). Here, "compatible" means the pointcloud fields start with XYZI in this order. Multiple data types, including uint8 and float, are supported for the datatype of the intensity field.