What problem does it solve? Working with raw 3D point clouds in ROS 2 requires understanding the packed binary PointCloud2 format and configuring the right PCL filters, which is error-prone without a reference. This Skill provides the message structure, filter configurations, and projection setup needed to process point clouds for navigation. ## Core Features & Use Cases - PointCloud2 Format Reference: Explains the binary buffer layout, common field arrangements (XYZ, XYZRGB, XYZI), and how to read points in Python with sensor_msgs_py or in C++ with pcl_conversions. - PCL Filter Configurations: Ready-to-use YAML parameters for VoxelGrid downsampling, PassThrough cropping, StatisticalOutlierRemoval denoising, and CropBox robot-body removal. - 3D-to-2D Projection: Configures pointcloud_to_laserscan to convert depth camera clouds into virtual LaserScan messages consumable by Nav2 costmaps. - Use Case: A robot with a depth camera needs obstacle detection in Nav2. Use this Skill to downsample the cloud, strip the floor and robot body, then project it to a LaserScan registered as a costmap observation source. ## Quick Start Ask the assistant to configure a pointcloud_to_laserscan pipeline that filters a depth camera cloud and feeds it into the Nav2 costmap.
