What problem does it solve? Integrating depth cameras (Intel RealSense, OAK-D, Azure Kinect) into a ROS2/Nav2 robot stack requires choosing between point cloud and virtual laser scan paths, wiring correct topic remappings, and handling optical frame conventions—mistakes here silently break costmap obstacle detection. ## Core Features & Use Cases - Point Cloud Generation: Set up depth_image_proc composable nodes (point_cloud_xyz, point_cloud_xyzrgb) to convert depth images and CameraInfo into PointCloud2 topics. - Virtual Laser Scan Conversion: Configure depthimage_to_laserscan to produce lightweight 2D LaserScan data directly from depth images with tunable scan height and range limits. - Costmap Integration: Add the depth-derived scan as an observation source so Nav2 costmaps detect low obstacles (table legs, furniture) that a 2D LIDAR misses. - Use Case: A mobile robot with a RealSense camera needs to avoid chair legs below its LIDAR plane; use this Skill to launch a depthimage_to_laserscan node and register its output as a costmap observation source. ## Quick Start Set up a depth camera pipeline that converts my RealSense depth image into a virtual laser scan and adds it as an observation source to my Nav2 costmap.
