Python API

pywavemap

A fast, efficient and accurate multi-resolution, multi-sensor 3D occupancy mapping framework.

class pywavemap.Map

Base class for wavemap maps.

add_to_cell_value: Increment the value of the map at a given index.

clear: Erase all cells in the map.

create = <nanobind.nb_func object>

property empty: Whether the map is empty.

get_cell_value: Query the value of the map at a given index.

interpolate: Query the map’s value at a point, using the specified interpolation mode.

load = <nanobind.nb_func object>

property max_index: Index of the maximum corner of the map’s Axis Aligned Bounding Box.

property max_log_odds: Upper threshold for the occupancy values stored in the map, in log-odds.

property memory_usage: The amount of memory used by the map, in bytes.

property min_cell_width: Maximum map resolution, set as width of smallest cell it can represent.

property min_index: Index of the minimum corner of the map’s Axis Aligned Bounding Box.

property min_log_odds: Lower threshold for the occupancy values stored in the map, in log-odds.

prune: Free up memory by pruning nodes that are no longer needed. Note that this pruning operation is lossless and does not alter the estimated occupancy posterior.

prune_smart: Similar to prune(), but avoids de-allocating nodes that were recently updated and will likely be used again in the near future.

set_cell_value

property size: The number of cells or nodes in the map, for fixed or multi-resolution maps, respectively.

store: Store a wavemap map as a .wvmp file.

threshold: Threshold the occupancy values of all cells in the map to stay within the range specified by its min_log_odds and max_log_odds.

property tree_height: Height of the octree used to store the map. Note that this value is only defined for multi-resolution maps.

class pywavemap.HashedWaveletOctree

Bases: Map

A class that stores maps using hashed wavelet octrees.

get_cell_value

Overloaded function.

get_cell_value(self, index: numpy.ndarray[dtype=int32, shape=(3), order='C']) -> float

Query the value of the map at a given index.

get_cell_value(self, node_index: pywavemap._pywavemap_bindings.OctreeIndex) -> float

Query the value of the map at a given octree node index.

get_cell_values

Overloaded function.

get_cell_values(self, index_list: ndarray[dtype=int32, shape=(*, 3), device='cpu']) -> numpy.ndarray[dtype=float32]

Query the map at the given indices, provided as a matrix with one (x, y, z) index per row.

get_cell_values(self, node_index_list: ndarray[dtype=int32, shape=(*, 4), device='cpu']) -> numpy.ndarray[dtype=float32]

Query the map at the given node indices, provided as a matrix with one (height, x, y, z) node index per row.

interpolate

Overloaded function.

interpolate(self, position: numpy.ndarray[dtype=float32, shape=(3), order='C'], mode: pywavemap._pywavemap_bindings.InterpolationMode = InterpolationMode.TRILINEAR) -> float

Query the map’s value at a point, using the specified interpolation mode.

interpolate(self, position_list: ndarray[dtype=float32, shape=(*, 3), device='cpu'], mode: pywavemap._pywavemap_bindings.InterpolationMode = InterpolationMode.TRILINEAR) -> numpy.ndarray[dtype=float32]

Query the map’s value at the given points, using the specified interpolation mode.

class pywavemap.HashedChunkedWaveletOctree

Bases: Map

A class that stores maps using hashed chunked wavelet octrees.

get_cell_value

Overloaded function.

get_cell_value(self, index: numpy.ndarray[dtype=int32, shape=(3), order='C']) -> float

Query the value of the map at a given index.

get_cell_value(self, node_index: pywavemap._pywavemap_bindings.OctreeIndex) -> float

Query the value of the map at a given octree node index.

interpolate: Query the map’s value at a point, using the specified interpolation mode.

class pywavemap.InterpolationMode(value)

NEAREST = 0

TRILINEAR = 1

class pywavemap.OctreeIndex

A class representing indices of octree nodes.

compute_child_index: Compute the index of the node’s n-th child, where n ranges from 0 to 7.

compute_parent_index

Overloaded function.

compute_parent_index(self) -> pywavemap._pywavemap_bindings.OctreeIndex

Compute the index of the node’s direct parent.

compute_parent_index(self, parent_height: int) -> pywavemap._pywavemap_bindings.OctreeIndex

Compute the index of the node’s parent (or ancestor) at parent_height.

property height: The node’s resolution level in the octree. A height of 0 corresponds to the map’s maximum resolution. In a fully allocated tree, all leaf nodes are at height 0. Increasing the height by 1 doubles the node size along each dimension. The root node corresponds to the map’s lowest resolution, and the root node’s height matches the configured tree height.

property position: The node’s XYZ position in the octree’s grid at the resolution level set by height.

class pywavemap.Rotation

A class representing rotations in 3D space.

inverse: Compute the rotation’s inverse.

class pywavemap.Pose

A class representing poses in 3D space.

inverse: Compute the pose’s inverse.

class pywavemap.Pointcloud: A class to store pointclouds.

class pywavemap.PosedPointcloud: A class to store pointclouds with an associated pose.

class pywavemap.Image: A class to store depth images.

class pywavemap.PosedImage: A class to store depth images with an associated pose.

class pywavemap.Pipeline

A class to build pipelines of measurement integrators and map operations.

add_integrator: Create and register a new integrator

add_operation: Create and register a new map operation.

clear: Deregister all the pipeline’s measurement integrators and map operations.

clear_integrators: Deregister all integrators.

clear_operations: Deregister all map operations

has_integrator: Returns true if an integrator with the given name has been registered.

remove_integrator: Deregister the integrator with the given name. Returns true if it existed.

run_integrators

Overloaded function.

run_integrators(self, integrator_names: typing.Sequence[str], posed_pointcloud: pywavemap._pywavemap_bindings.PosedPointcloud) -> bool

Integrate a given pointcloud.

run_integrators(self, integrator_names: typing.Sequence[str], posed_image: pywavemap._pywavemap_bindings.PosedImage) -> bool

Integrate a given depth image.

run_operations: Run the map operations.

run_pipeline

Overloaded function.

run_pipeline(self, integrator_names: typing.Sequence[str], posed_pointcloud: pywavemap._pywavemap_bindings.PosedPointcloud) -> bool

Integrate a given pointcloud, then run the map operations.

run_pipeline(self, integrator_names: typing.Sequence[str], posed_image: pywavemap._pywavemap_bindings.PosedImage) -> bool

Integrate a given depth image, then run the map operations.

convert

Submodule with common conversion functions for wavemap index types.

convert.cell_width_to_height(cell_width: float, min_cell_width: float) → int

Compute the minimum node height (resolution level) required to reacha given node width.

param cell_width:

The desired node width.

param min_cell_width:

The grid resolution at height 0 (max map resolution).

convert.height_to_cell_width(min_cell_width: float, height: int) → float

Compute the node width at a given height.

param min_cell_width:

The grid resolution at height 0 (max map resolution).

param height:

The desired height (resolution level) of the node index.

convert.point_to_nearest_index(point: numpy.ndarray[dtype=float32, shape=(3), order='C'], cell_width: float) → numpy.ndarray[dtype=int32, shape=(3), order='C']: Compute the nearest index to a point on a grid with a given cell width.

convert.point_to_node_index(point: numpy.ndarray[dtype=float32, shape=(3), order='C'], min_cell_width: float, height: int) → pywavemap._pywavemap_bindings.OctreeIndex

Compute the index of a node containing a given point.

param min_cell_width:

The grid resolution at height 0 (max map resolution).

param height:

The desired height (resolution level) of the node index.

logging

Submodule to configure wavemap’s logging system.

logging.set_level(level: str = 'info') → None: Set pywavemap’s logging level.

logging.enable_prefix(enable: bool = False) → None: Whether to prefix log messages with timestamps and line numbers.

param

Submodule for wavemap’s config system.

class pywavemap.param.Value: A class that holds parameter values. Note that one Value can hold a primitive type, a list of Values, or a dictionary of Values. One Value can therefore hold the information needed to initialize an entire config, or even a hierarchy of nested configs.

Typedef wavemap::VectorDistanceField