What problem does it solve? 2D costmap layers project all sensor data onto a plane, so obstacles detected by a high-mounted depth camera get erased by a low lidar's clearing rays, and overhanging obstacles like shelves or table edges are missed entirely. This Skill guides correct configuration of the Nav2 VoxelLayer, which maintains a true 3D voxel grid with volumetric raycasting. ## Core Features & Use Cases - 3D Voxel Grid Configuration: Set z_voxels, z_resolution, origin_z, and mark_threshold to define the monitored vertical volume and control 3D-to-2D projection sensitivity. - Multi-Height Sensor Fusion: Combine a low 2D lidar and a high depth camera in one voxel layer so clearing rays only affect voxels at the ray's actual height. - Debugging with Voxel Map Visualization: Enable publish_voxel_map to inspect marked voxels in RViz and diagnose ground-hit false positives or stuck obstacles. - Use Case: A mobile robot with a lidar at 15cm and a depth camera at 80cm must track table-top obstacles; the VoxelLayer keeps camera marks in upper voxels intact even when the lidar raytraces through the same (x, y) column below. ## Quick Start Ask the AI to configure a Nav2 local costmap voxel layer that fuses a LaserScan lidar and a PointCloud2 depth camera with proper z_voxels and height filtering.
