Road Occupancy Processor

This package generates the occupancy grid indicating the status of the road. It uses the point cloud from a filtered sensor and the ADAS Map data.

The occupancy grid can be conceptualized as a one dimensional 8-bit depth bitmap.

This package publishes a GridMap/OccupancyGrid with four different possible values:

• UNKNOWN
• FREE
• NO ROAD
• OCCUPIED

These values can be set using parameters, as described in the Configuration Parameters section below.

Extra requirements for compilation

• GridMap (http://wiki.ros.org/grid_map)

Prerequisites

• Localization working (TF between velodyne and world)

  • PCD Map (map)
  • TF for the Map (tf map->world)
  • NDT matching (tf world-> base_link)
  • Base link to Localizer (tf base_link -> velodyne)

• Ground filtering method publishing /points_ground and /points_no_ground. Autoware has three available:

  • ray_ground_filter
  • ring_ground_filter
  • euclidean_cluster with Planar Ground removal

• wayarea2gridmap node from Semantics in the Computing Tab

  • Vector Map in /vector_map (also known as ADAS Map)
    • VectorMap must contain the definition of road areas (way_areas)

Data subscription

This node subscribes to:

• Ground filtered in /points_ground (sensor_msgs::PointCloud)
• Obstacle points in /points_no_ground (sensor_msgs::PointCloud)
• GridMap containing the road processed in /grid_map_wayarea (grid_map::GridMap)

Data publishing

• gridmap_road_status publishes the native format of the grid_map package. This can be used to extend the functionalities of this package.
• occupancy_road_status publishes the native ros nav_msgs::OccupancyGrid message.

How to run

From a sourced terminal in the workspace: roslaunch road_occupancy_processor road_occupancy_processor.launch

Quick Setup

This is only a quick guide, each node must be properly configured

1. Load PointCloud Map (Map tab)
2. Load VectorMap with wayareas available (Map tab)
3. Load TF between map and world frames (Map tab)
4. Load TF between base_link and Localizer (Setup tab)
5. Launch voxel_grid_filter (Sensing tab/Points Downsampler)
6. Launch ray_ground_filter (Sensing tab/Points Preprocessor)
7. Launch ndt_matching (Computing tab/Localization)
8. Initialize NDT pose
9. Launch wayarea2grid (Computing tab/Localization)
10. Launch road_occupancy_processor (Computing tab/Localization)

Configuration parameters

• points_ground_src (default="points_ground") defines the PointCloud source topic containing only the ground points.
• points_no_ground_src (default="points_no_ground") defines the PointCloud source topic containing only the obstacle points.
• wayarea_topic_src (default=grid_map_wayarea) defines the name of the topic containing the grid_map_msgs:GridMap containing the road areas.
• wayarea_layer_name (default="wayarea") defines the name of the layer in the topic wayarea_topic_src that contains the road areas.
• output_layer_name (default=road_status) defines name of the output layer in the published GridMap object. If running several instances, each vehicle one can publish a different layer and later add them.
• road_unknown_value (default="128") indicates the value to fill in the occupancy grid when a cell is UNKNOWN.
• road_free_value (default="75") indicates the value to fill in the occupancy grid when a cell is FREE. Should be a number between 0-255.
• road_occupied_value (default="0") indicates the value to fill in the occupancy grid when a cell is OCCUPIED. Should be a number between 0-255.
• no_road_value (default="255") indicates the value to fill in the occupancy grid when a cell is NO ROAD. Should be a number between 0-255.

Coordinate Frame

The occupancy grid is published in the same coordinate frame as the input GridMap from /grid_map_wayarea.

Color representation using default values

Black color represents areas not defined as NO ROAD in the vector map.

Dark gray indicates UNKNOWN areas as detected by this node.

Light gray means the area is FREE.

White is used for OCCUPIED areas.