knb / hypergrid

Hypergrid is a ROS package for building local maps from multiple sensors in the blink of an eye

Hypergrid

Hypergrid is a ROS package for building local maps from multiple sensors in the blink of an eye. Hypergrid relies on Arrayfire, in order to parallelize the computation.

Dependencies

• Arrayfire

What's inside

Library

Hypergrid is packaged with catkin and provides several libraries to work with grid maps and to convert the raw sensor data to this format.

Costmap Plugin

The main contribution of this package is a costmap_2d plugin, meant to replace the default obstacle_layer inside the local_costmap of move_base. The typical local_costmap_params.yaml configuration file can be configured as follows:

local_costmap:
  
  # ...
  
  plugins:
   - {name: hypergrid_layer,          type: "hypergrid::HypergridLayer"}
  
  hypergrid_layer:
    # LaserScan input topics
    laser_topics: ["icab1/scan"]
    # LiDAR PointCloud2 input topics
    lidar_topics: ["icab1/velodyne_points"]
    # Frame of the output local map
    map_frame_id: icab1/base_footprint
    # Map size (in meters)
    width: 100
    height: 100
    # Map resolution (in meters / cell)
    cell_size: 0.1
    # Remove floor points in pointcloud with heightmap algorithm
    floor_filter: true
    # Heightmap algorithm parameters
    heightmap_threshold: 0.15
    heightmap_cell_size: 0.25
    # Remove all points higher than max_height
    max_height: 2.5
    # Remove all points inside the vehicle box
    # All points with |x| < vehicle_box_size or |y| < vehicle_box_size will be filtered.
    vehicle_box_size: 2.0
    # Enable debug console output
    DEBUG: true

Nodes

In addition to the library and the costmap_2d plugin, Hypergrid also includes stand-alone conversion nodes, which subscribe to the sensor messages and publish the output instantaneous local map.

• laser_to_gridmap: Conversion node from LaserScan msgs to OccupancyGrid maps.

• pcl_to_gridmap: Conversion node from LiDAR PointCloud2 msgs to OccupancyGrid maps.

The launch and configuration files are inside the launch and config directories, respectively.