A Pose is a tool to interpret tensors of float as SE3 poses Parameters ---------- config : dict A dictionary with the config

A SphericalProjector projects a pointcloud in a spherical image Parameters ---------- up_fov : float The field of view upwar

The Configuration for the Point-To-Plane ICP based Iterative Least Square estimation of the pose

A dataclass wrapper for a pct.DatasetOptions The fields of the dataclass are programmatically defined from the attributes of a pct.DatasetOpt

Gauss Newton algorithm for Least Square minimization

Posenet Module

An util object which progressively computes the mean over logged values

2D Feature based registration which estimates the planar motion (x, y, yaw) Only relevant for a sensor having "mainly 2D" motion, and can ser

Point-to-Plane Cost function

Object which performs interpolation of poses using timestamps Poses and corresponding key timestamps are passed to the constructor. The Poses

Config for a Rosbag Dataset

A SLAMRunner runs a LiDAR SLAM algorithm on a set of pytorch datasets, And if the ground truth is present, it evaluates the performance of th

Sequence of the UrbanLoco Dataset wrapping a Rosbag Note: As the dataset is a rosbag dataset, random access is not authorized

A Dataset which concatenates data into for a small window of frames Takes a list of Datasets, each corresponding to a sequence. The data

A Module which exponentially weights different losses during training It holds parameters weigh the different losses. The weights change

Ford Campus Sequence

A PoseGraph backend which maintains and performs global optimization Args: Whether to initialize the world coordinates at the fir

The Configuration for the Point-To-Plane ICP based Iterative Least Square estimation of the pose

Dataset for a Sequence of KITTI-360 lidar dataset Attributes: kitti360_root_dir (str): The path to KITTI-360 data drive_id (

Dataset for a Sequence of KITTI odometry's benchmark Attributes: sequences_root_dir (str): The path to KITTI odometry benchmark's da

A Torch Dataset for a sequence of the NCLT Dataset see http://robots.engin.umich.edu/nclt for the link to the Dataset's main page TODO:

Dataset for a Sequence of the New Handheld College Dataset see https://ori-drs.github.io/newer-college-dataset/ Attributes: sequ

The configuration without initialization

An object which aggregrates the results of an Odometry benchmark

The Configuration for the Point-To-Plane ICP based Iterative Least Square estimation of the pose

Unsupervised / Supervised Trainer for the PoseNet prediction module

A Config for a PoseNetTrainer

PoseNet Prediction Config

A `Preprocessing` instance applies a sequence of `Filter`(s) a data_dict

The configuration for `Preprocessing`

A ResNet Encoder consist of the 4 layers of a ResNet which can be returned and decoded

A Dataset which wraps a RosBag Note: The dataset can only read data sequentially, and will raise an error when two calls are not

A SLAM Algorithm for Point Cloud data (typically LiDAR Data) A SLAM of pyLIDAR-SLAM consists of four modules - Motion Initialization

The configuration dataclass

A Window datasets wraps a dataset_config, and limit it to a defined window

A Visualizer displays images tensors in OpenCV windows Parameters ---------- channels : list of str The

Point-to-Plane Loss Module

Supervised Loss Module

An abstract Trainer class is the backbone for training deep learning modules Each ATrainer child classes defines a prediction Module and a l

The configuration dataclass for a Trainer

An abstract class for the backend of a SLAM algorithm Given a set of trajectory constraints, the backend performs a global optimization Takin

An abstract class for the Config for a Backend

A configuration object read from a yaml conf

Configuration for a dataset proposed in CT_ICP

Dataset for a Sequence defined in CT_ICP Datasets See https://github.com/jedeschaud/ct_icp for more details Attributes:

An Odometry Algorithm which updates the poses by taking into account the distortion of the frame The algorithm uses the wrapping pyt

The configuration without initialization

A Constant Velocity model for initialization (returns the last registered relative pose at each time step)

ResNet basic block where the ReLU activation is replaced by a Custom activation

ResNet Bottleneck block where the ReLU activation is replaced by a Custom activation

The different datasets covered by the dataset_config configuration A configuration must have the field dataset_config pointing to one of thes

A DatasetConfig contains the configuration values used to define a DatasetConfiguration

A DatasetConfiguration is the configuration for the construction of pytorch Datasets

Distort a frame using the estimated initial motion

A Filter Config a distortion of a frame

Config for Elevation Image feature based 2D alignment

Configuration for a ElevationImageLoopClosure Algorithm

Initialize motion by resolving a planar motion registration

An Implementation of a Loop Detection and Estimation Algorithm

Filters registered

A Filter on the input slam data

A Configuration for a filter

A Configuration object read from a yaml conf

Configuration for Ford Dataset

Configuration for a Gauss Newton based Point-to-Point rigid alignment

A GaussNewton Point-To-Plane rigid alignment method, Which minimizes the Point-To-Plane distance

Configuration for a Gauss-Newton based Point-to-Plane rigid alignment

A GaussNewton Point-To-Point rigid alignment method, Which minimizes the Point-To-Point distance

Distort a frame using the estimated initial motion

A Filter Config for the sampling of a frame

OdometryAlgorithm based on the ICP-registration

A Convenient enum to load the Algorithm from a config dictionary

Scan registration method using feature based Image Alignment.

The initialization module provides for each frame a prior estimate of the relative motion Each child class adds an [4, 4] numpy ndarray of the re

The Initialization Config for registration based SLAM

A configuration object read from a yaml conf

Configuration for KITTI-360 dataset see http://www.cvlibs.net/datasets/kitti-360/

A configuration object read from a yaml conf

Configuration for KITTI's dataset odometry benchmark see http://www.cvlibs.net/datasets/kitti/eval_odometry.php

A Local Map which computes neighborhood by projective association The local map keeps in memory the last N processed Frames

A KdTree-based Local Map

Convenient Enum to load LocalMap from configuration

An Abstract class for Least Square Minimization