robot_body_filter
A ROS filter that removes points belonging to a robot's own body from laser scans and point clouds.
What is robot_body_filter?
robot_body_filter is a ROS-based filter that removes points corresponding to a robot's own body from laser scans and point clouds. It solves the problem of self-occlusion in robotic perception, where the robot's body appears in its sensor data, corrupting tasks like mapping, obstacle detection, and navigation. The filter uses the robot's URDF model and TF transforms to identify and filter out these points through containment and ray-tracing tests.
ROS developers and roboticists working on perception systems for mobile manipulators, autonomous vehicles, or any robot with onboard sensors that can see its own body. It is particularly useful for those using lidars, RGB-D cameras, or laser scanners in dynamic environments.
Developers choose robot_body_filter for its improved accuracy over predecessors like robot_self_filter, its support for both organized and unorganized point clouds, and its extensive configurability for different sensor modalities and robot motions. The ability to compute bounding shapes and provide detailed debugging outputs makes it a versatile tool for robust perception pipelines.
Overview
Filters the robot's body out of point clouds and laser scans.
Use Cases
Best For
- Removing robot arm self-occlusions from point clouds in manipulation tasks
- Cleaning lidar scans from mobile robots to improve SLAM accuracy
- Filtering out sensor mounts and robot body parts from RGB-D camera data
- Generating bounding volumes of the robot for collision checking or visualization
- Debugging perception pipelines by visualizing which points are classified as robot body
- Handling sensors on moving parts with point-by-point temporal interpolation
Not Ideal For
- Projects with strict real-time requirements on resource-constrained embedded hardware
- Teams seeking a drop-in filter without extensive parameter tuning and TF frame configuration
- Non-ROS robotics frameworks or custom perception pipelines outside the ROS ecosystem
- Applications where only basic distance-based clipping is needed and containment/shadow tests are unnecessary
Pros & Cons
Pros
Combines clipping, containment tests, and shadow tests to comprehensively remove robot body points, as described in the general overview for robust data cleaning.
Supports both all-at-once and point-by-point acquisition modes with proper temporal interpolation, essential for sensors like spinning lidars, per the data acquisition modes section.
Publishes visualization markers and point clouds for each filtering class (INSIDE, CLIP, SHADOW), aiding in setup and verification as highlighted in the debug operation section.
Allows parameters to limit processed links, shadow test distance, and model update frequency to balance accuracy and CPU usage, detailed in the performance tips.
Cons
With over 50 parameters and nuanced TF frame settings, initial setup requires significant effort, as admitted in the tutorial and examples folder for inspiration.
Shadow tests involve ray-tracing and can be slow, especially with complex mesh models, necessitating performance optimizations that may reduce accuracy, per the performance tips.
The oriented bounding box computation is described as 'very bad conditioned' and 'non-satisfying' in the README, limiting its usefulness for precise volume estimation.
Frequently Asked Questions
Related Projects
GitHub repositoryOpen3D: A Modern Library for 3D Data Processing
DracoDraco is a library for compressing and decompressing 3D geometric meshes and point clouds. It is intended to improve the storage and transmission of 3D graphics.
mmdetection3dOpenMMLab's next-generation platform for general 3D object detection.
PCDetOpenPCDet Toolbox for LiDAR-based 3D Object Detection.
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.