Maps

While frames relate cartesian frames to each other, maps are the objects in which the actual information about the environment is stored.

All maps in envire are of the base class Layer. The layer class provides information on the dirty status of a map, so if a map has been modified or not, and can store generic metadata. See the API doc for a class hierarchy of the Layer.

Layers itself do not contain any spatial information, so they can in principle store both cartesian as well as non-cartesian (topological) maps. The derived class CartesianMap does provide the ability to associate a layer with a frame. This is done through either

env->setFrameNode( map, frameNode );

alternatively you can use the convenience method on the map directly, if it is already attached.

env->attachItem( map );
map->setFrameNode( frameNode );

Each CartesianMap is associated with a frame, and the information it has is given in that frame. Further down the inheritance chain, there is the Map class, which is templated for the map dimension. Currently only Map<2> and Map<3> are used.

Metadata

Each Layer can have associated metadata, which may or may not be used by the processing algorithms (operators).

if( map->hasData( key ) )
{
    DataType &d = map->getData<DataType>( key );

    ... // do stuff on d
}

a call to getData<Type> will create the metadata if it is not yet created. Metadata on maps is used to specify optional information that can be available for specific maps, but which might not be available for all forms of that particular map. This was chosen instead of inheritance to simplify handling of different permutations of available data.

Some maps may provide additional features for metadata, like e.g. in a Pointcloud. Here vertices are stored as a vector of Vector3 objects. However, these vertices may also have associated normals.

if( pc->hasData( Pointcloud::VERTEX_NORMALS ) )
{
    std::vector<Eigen::Vector3d> normals =
	pc->getVertexData<Eigen::Vector3>(
		Pointcloud::VERTEX_NORMALS );

    // both vectors should be of the same length
    assert( pc->vertices.size() == normals.size() );

    // perform a calculation on vertex normal pairs
    for( size_t n=0; n<pc->vertices.size(); n++ )
	doCalc( pc->vertices[n], normals[n] );
}

Examples of Maps

There is a large number of maps available in envire, the best is to have a look at the API docs. In principle, one can distinguish between dense and sparse maps. E.g. a Pointcloud is a sparse map, while a Grid is a dense 2d map.