dynamic_cloud_detector

Overview

The white and red point clouds are determined to be static and dynamic, respectively.

The color of the map indicates occupancy; the blacker the area, the higher the occupancy.

Enviornment

• Ubuntu 16.04 or 18.04
• ROS Kinetic or Melodic

Requirement

• PCL 1.8
• velodyne_height_map(https://github.com/jack-oquin/velodyne_height_map)

Install and Build

cd catkin_workspace/src
git clone https://github.com/amslabtech/dynamic_cloud_detector.git
cd ..
catkin_make

Nodes

dynamic_cloud_detector

Published topics

• /cloud/static (sensor_msgs/PointCloud2)
  • point cloud on high-occupancy grid
• /cloud/dynamic (sensor_msgs/PointCloud2)
  • point cloud on low-occupancy grid
• ~/occupancy_grid (nav_msgs/OccupancyGrid)
  • for visualization

Subscribed topics

• /velodyne_obstacles (sensor_msgs/PointCloud2)
  • point cloud without ground
• /odom/complement (nav_msgs/Odometry)
  • odometry complemented by IMU. raw odometry or estimated pose is also allowed

Parameters

• RESOLUTION
  • resolution of occupancy grid map (default: 0.25[m])
• WIDTH
  • widht and height of occupancy grid map (default: 40[m])
• OCCUPANCY_THRESHOLD
  • the grid cell whose occupancy is lower than this threshold is considered as dynamic (default: 0.2)
• BEAM_NUM
  • this node uses beam model for searching free space (default: 720)
• BUFFER_SIZE
  • this parameter defines the buffer size for point cloud and odometry (default: 40)
• SKIP_SCAN_FLAG
  • if this flag is true, this node adds new data to buffer once in twice (default: false)

How to Use

roslaunch dynamic_cloud_detector dynamic_cloud_detector.launch