open3d.t.pipelines.slam.Model¶

class open3d.t.pipelines.slam.Model¶

Volumetric model for Dense SLAM.

__init__(*args, **kwargs)¶

Overloaded function.

__init__(self)
__init__(self, voxel_size, block_resolution=16, block_count: int = 10000, transformation=(with default value), , device=CUDA:0)
Constructor of a TSDFVoxelGrid

Parameters

voxel_size (float) – The voxel size of the volume in meters.
(int (block_resolution) – int = 10000): Resolution of local dense voxel blocks. By default 16 is used to create 16^3 voxel blocks.
optional – int = 10000): Resolution of local dense voxel blocks. By default 16 is used to create 16^3 voxel blocks.
default=16 – int = 10000): Resolution of local dense voxel blocks. By default 16 is used to create 16^3 voxel blocks.
block_count – int = 10000): Resolution of local dense voxel blocks. By default 16 is used to create 16^3 voxel blocks.
transformation (open3d.core.Tensor, optional) –
A 4x4 3D transformation matrix. Default value:

[[1.0 0.0 0.0 0.0], [0.0 1.0 0.0 0.0], [0.0 0.0 1.0 0.0], [0.0 0.0 0.0 1.0]] Tensor[shape={4, 4}, stride={4, 1}, Float64

()
(open3d.core.Device (device) – 0): The CPU or CUDA device used for the object.
optional – 0): The CPU or CUDA device used for the object.
default=CUDA – 0): The CPU or CUDA device used for the object.

extract_pointcloud(self, estimated_number=- 1, weight_threshold=3.0)¶

Extract point cloud from the volumetric model.

Parameters

estimated_number (int, optional, default=-1) – Estimated number of surface points. Use -1 if no estimation is available.
weight_threshold (float, optional, default=3.0) – Weight threshold to filter outlier voxel blocks.

Returns

open3d.t.geometry.PointCloud

extract_trianglemesh(self, estimated_number=- 1, weight_threshold=3.0)¶

Extract triangle mesh from the volumetric model.

Parameters

estimated_number (int, optional, default=-1) – Estimated number of surface points. Use -1 if no estimation is available.
weight_threshold (float, optional, default=3.0) – Weight threshold to filter outlier voxel blocks.

Returns

open3d.t.geometry.TriangleMesh

get_current_frame_pose(self: open3d.cpu.pybind.t.pipelines.slam.Model) → open3d.cpu.pybind.core.Tensor¶

get_hashmap(self: open3d.cpu.pybind.t.pipelines.slam.Model) → open3d.cpu.pybind.core.HashMap¶: Get the underlying hash map from 3D coordinates to voxel blocks.

integrate(self, input_frame, depth_scale=1000.0, depth_max=3.0)¶

Integrate an input frame to a volume.

Parameters

input_frame (open3d.t.pipelines.slam.Frame) – The frame that contains raw depth and optionally images with the same size from the input.
depth_scale (float, optional, default=1000.0) – The scale factor to convert raw depth into meters.
depth_max (float, optional, default=3.0) – The max clipping depth to filter noisy observations too far.

Returns

None

synthesize_model_frame(self, model_frame, depth_scale=1000.0, depth_min=0.1, depth_max=3.0, enable_color=False)¶

Synthesize frame from the volumetric model using ray casting.

Parameters

model_frame (open3d.t.pipelines.slam.Frame) – The frame that contains ray casted depth and optionally color from the volumetric model.
depth_scale (float, optional, default=1000.0) – The scale factor to convert raw depth into meters.
depth_min (float, optional, default=0.1) – The min clipping depth.
depth_max (float, optional, default=3.0) – The max clipping depth to filter noisy observations too far.
enable_color (bool, optional, default=False) –

Returns

None

track_frame_to_model(self, input_frame, model_frame, depth_scale=1000.0, depth_max=3.0, depth_diff=0.07)¶

Track input frame against raycasted frame from model.

Parameters

input_frame (open3d.t.pipelines.slam.Frame) – The frame that contains raw depth and optionally images with the same size from the input.
model_frame (open3d.t.pipelines.slam.Frame) – The frame that contains ray casted depth and optionally color from the volumetric model.
depth_scale (float, optional, default=1000.0) – The scale factor to convert raw depth into meters.
depth_max (float, optional, default=3.0) – The max clipping depth to filter noisy observations too far.
depth_diff (float, optional, default=0.07) –

Returns

open3d.t.pipelines.odometry.OdometryResult

update_frame_pose(self: open3d.cpu.pybind.t.pipelines.slam.Model, arg0: int, arg1: open3d.cpu.pybind.core.Tensor) → None¶

property transformation_frame_to_world¶: Get the 4x4 transformation matrix from the current frame to the world frame.