interfaces.md

VineSLAM ROS architecture

---

ROS nodes

• slam_node - Simultaneous Localization and Mapping node
• localization_node - Localization-only node. Loads a map at the initialization.
• mapping_node - Mapping with known poses node.

Subscribed topics

• "/detections_topic" (vision_msgs::Detection2DArray) - the object detections
• "/features_topic" (vineslam_msgs::FeatureArray) - the object detections
• "/scan_topic" (sensor_msgs::PointCloud2) - the object detections
• "/odom_topic" (nav_msgs::Odometry) - the robot wheel Odometry
• "/gps_topic" (sensor_msgs::NavSatFix) - the GPS topic (optional)

Published topics (main)

• /vineslam/pose (geometry_msgs::PoseStamped) - the 6-DoF robot pose
• /vineslam/poses (geometry_msgs::PoseArray) - the 6-DoF particles pose
• /vineslam/map2D (visualization_msgs::MarkerArray) - the 2D semantic map
• /vineslam/map3D/SURF (sensor_msgs::PointCloud2) - the 3D SURF-based map
• /vineslam/map3D/corners (sensor_msgs::PointCloud2) - the 3D PointCloud corner feature map
• /vineslam/map3D/planars (sensor_msgs::PointCloud2) - the 3D PointCloud planar features map
• /vineslam/map3D/planes (visualization_msgs::MarkerArray) - the 3D PointCloud plane features map
• /vineslam/elevationMap (visualization_msgs::MarkerArray) - the ground elevation map

Transform broadcast

• "/odom" to world_frame_id (map).

Parameters

To tune the system parameters, you can change them in ./vineslam_ros/config/setup_slam.yaml.

All the parameters are set to their default values, so you can run the system witouth changing them.

• system:
  • robot_model - [agrob/rocha] robot model to use (will have impact in the tf broadcasting).
  • world_frame_id - name of the world reference frame.
  • use_semantic_features - [True/False] whether to use or not the 2D semantic feature map
  • use_lidar_features - [True/False] whether to use or not the 3D LiDAR features
  • use_image_features - [True/False] whether to use or not the 3D image features
  • use_gps - [True/False] whether to use or not the GPS on 6-DoF Localization
• camera_info:
  • fx - horizontal focal length (pixels)
  • cx - x coordinate of image principal point (pixels)
• multilayer-mapping:
  • grid_map:
    • origin:
      • x - x coordinate of grid map origin
      • y - y coordinate of grid map origin
      • z - z coordinate of grid map origin
    • width - map width (meters)
    • length - map lenght (meters)
    • height - map height (meters)
    • resolution - map resolution (meters)
    • save_map - [True/False] whether to save or not the map when the system ends
    • output_file - name of the map output file (.xml)
    • input_file - name of the map input file (.xml) if using localization-only
• pf:
  • n_particles - number of particles of the particle filter
  • sigma_xx - motion model xx constant (meters)
  • sigma_yy - motion model yy constant (meters)
  • sigma_zz - motion model zz constant (meters)
  • sigma_RR - motion model RR constant (radians)
  • sigma_PP - motion model PP constant (radians)
  • sigma_RR - motion model YY constant (radians)
  • k_clusters - number of clusters to consider at the end of the particle filter