Realsense-RGBD-Camera-Position-Detection

The program is based on ros_pcl and pcl, filter and cluster the raw pointcloud data to get object position at 30Hz.

How to use it?

1.install librealsense2 and realsense ros

2.verify that the packages are installed successfully

roslaunch realsense2_camera rs_camera.launch

3.add the files in repository to realsense2_camera package. If the file already exits (eg.CMakeLists.txt), please substitute it.

4.run the program

launch the node to publish raw pointcloud

roslaunch realsense2_camera demo_poincloud.launch

launch the nodelets to preprocess pointcloud

roslaunch realsense2_camera pcl_preprocess.launch

run the node to clustering pointcloud

rosrun realsense2_camera detection

5.detected objects information is published in topic /objectlist as user-defined ros message

Parameters

pcl_preprocess.launch

filter_limit_min: the minimum range to keep points. z-direction points outward from the camera.

filter_limit_max: the maximum range to keep points.

leaf_size: cube side length to down sampling.

distance_threshold: the threshold of plane thickness.

detection.cpp

setClusterTolerance: epsilon, radius to search points.

setMinClusterSize: minmum number of points in one cluster.

setMaxClusterSize: maximum number of points in one cluster.

Topics

• /voxel_grid/output

point cloud filtered by passthrough filter and voxel grid filter

• /extract_plane/output

point clould filtered by RANSAC

• /cloud_cluster

clustered point cloud

Result

The printed message is the serial number of objects and its position.