Poisson Surface Reconstruction for LiDAR Odometry and Mapping

Overview

Poisson Surface Reconstruction for LiDAR Odometry and Mapping

Surfels TSDF Our Approach
suma tsdf puma

Table: Qualitative comparison between the different mapping techniques for sequence 00 of the KITTI odometry benchmark.

This repository implements the algorithms described in our paper Poisson Surface Reconstruction for LiDAR Odometry and Mapping.

This is a LiDAR Odometry and Mapping pipeline that uses the Poisson Surface Reconstruction algorithm to build the map as a triangular mesh.

We propose a novel frame-to-mesh registration algorithm where we compute the poses of the vehicle by estimating the 6 degrees of freedom of the LiDAR. To achieve this, we project each scan to the triangular mesh by computing the ray-to-triangle intersections between each point in the input scan and the map mesh. We accelerate this ray-casting technique using a python wrapper of the Intel® Embree library.

The main application of our research is intended for autonomous driving vehicles.

Table of Contents

Running the code

NOTE: All the commands assume you are working on this shared workspace, therefore, first cd apps/ before running anything.

Requirements: Install docker

If you plan to use our docker container you only need to install docker and docker-compose.

If you don't want to use docker and install puma locally you might want to visit the Installation Instructions

Datasets

First, you need to indicate where are all your datasets, for doing so just:

export DATASETS=<full-path-to-datasets-location>

This env variable is shared between the docker container and your host system(in a read-only fashion).

So far we've only tested our approach on the KITTI Odometry benchmark dataset and the Mai city dataset. Both datasets are using a 64-beam Velodyne like LiDAR.

Building the apss docker container

This container is in charge of running the apss and needs to be built with your user and group id (so you can share files). Building this container is straightforward thanks to the provided Makefile:

make

If you want' to inspect the image you can get an interactive shell by running make run, but it's not mandatory.

Converting from .bin to .ply

All our apps use the PLY which is also binary but has much better support than just raw binary files. Therefore, you will need to convert all your data before running any of the apps available in this repo.

docker-compose run --rm apps bash -c '\
    ./data_conversion/bin2ply.py \
    --dataset $DATASETS/kitti-odometry/dataset/ \
    --out_dir ./data/kitti-odometry/ply/ \
    --sequence 07
    '

Please change the --dataset option to point to where you have the KITTI dataset.

Running the puma pipeline

Go grab a coffee/mate, this will take some time...

docker-compose run --rm apps bash -c '\
    ./pipelines/slam/puma_pipeline.py  \
    --dataset ./data/kitti-odometry/ply \
    --sequence 07 \
    --n_scans 40
    '

Inspecting the results

The pipelines/slam/puma_pipeline.py will generate 3 files on your host sytem:

results
├── kitti-odometry_07_depth_10_cropped_p2l_raycasting.ply # <- Generated Model
├── kitti-odometry_07_depth_10_cropped_p2l_raycasting.txt # <- Estimated poses
└── kitti-odometry_07_depth_10_cropped_p2l_raycasting.yml # <- Configuration

You can open the .ply with Open3D, Meshlab, CloudCompare, or the tool you like the most.

Where to go next

If you already installed puma then it's time to look for the standalone apps. These apps are executable command line interfaces (CLI) to interact with the core puma code:

├── data_conversion
│   ├── bin2bag.py
│   ├── kitti2ply.py
│   ├── ply2bin.py
│   └── ros2ply.py
├── pipelines
│   ├── mapping
│   │   ├── build_gt_cloud.py
│   │   ├── build_gt_mesh_incremental.py
│   │   └── build_gt_mesh.py
│   ├── odometry
│   │   ├── icp_frame_2_frame.py
│   │   ├── icp_frame_2_map.py
│   │   └── icp_frame_2_mesh.py
│   └── slam
│       └── puma_pipeline.py
└── run_poisson.py

All the apps should have an usable command line interface, so if you need help you only need to pass the --help flag to the app you wish to use. For example let's see the help message of the data conversion app bin2ply.py used above:

Usage: bin2ply.py [OPTIONS]

  Utility script to convert from the binary form found in the KITTI odometry
  dataset to .ply files. The intensity value for each measurement is encoded
  in the color channel of the output PointCloud.

  If a given sequence it's specified then it assumes you have a clean copy
  of the KITTI odometry benchmark, because it uses pykitti. If you only have
  a folder with just .bin files the script will most likely fail.

  If no sequence is specified then it blindly reads all the *.bin file in
  the specified dataset directory

Options:
  -d, --dataset PATH   Location of the KITTI dataset  [default:
                       /home/ivizzo/data/kitti-odometry/dataset/]

  -o, --out_dir PATH   Where to store the results  [default:
                       /home/ivizzo/data/kitti-odometry/ply/]

  -s, --sequence TEXT  Sequence number
  --use_intensity      Encode the intensity value in the color channel
  --help               Show this message and exit.

Citation

If you use this library for any academic work, please cite the original paper.

@inproceedings{vizzo2021icra,
author    = {I. Vizzo and X. Chen and N. Chebrolu and J. Behley and C. Stachniss},
title     = {{Poisson Surface Reconstruction for LiDAR Odometry and Mapping}},
booktitle = {Proc.~of the IEEE Intl.~Conf.~on Robotics \& Automation (ICRA)},
codeurl   = {https://github.com/PRBonn/puma/},
year      = 2021,
}
Owner
Photogrammetry & Robotics Bonn
Photogrammetry & Robotics Lab at the University of Bonn
Photogrammetry & Robotics Bonn
A Graph Neural Network Tool for Recovering Dense Sub-graphs in Random Dense Graphs.

PYGON A Graph Neural Network Tool for Recovering Dense Sub-graphs in Random Dense Graphs. Installation This code requires to install and run the graph

Yoram Louzoun's Lab 0 Jun 25, 2021
It is an open dataset for object detection in remote sensing images.

RSOD-Dataset It is an open dataset for object detection in remote sensing images. The dataset includes aircraft, oiltank, playground and overpass. The

136 Dec 08, 2022
Simple PyTorch hierarchical models.

A python package adding basic hierarchal networks in pytorch for classification tasks. It implements a simple hierarchal network structure based on feed-backward outputs.

Rajiv Sarvepalli 5 Mar 06, 2022
Unsupervised Pre-training for Person Re-identification (LUPerson)

LUPerson Unsupervised Pre-training for Person Re-identification (LUPerson). The repository is for our CVPR2021 paper Unsupervised Pre-training for Per

143 Dec 24, 2022
StudioGAN is a Pytorch library providing implementations of representative Generative Adversarial Networks (GANs) for conditional/unconditional image generation.

StudioGAN is a Pytorch library providing implementations of representative Generative Adversarial Networks (GANs) for conditional/unconditional image generation.

3k Jan 08, 2023
Code for technical report "An Improved Baseline for Sentence-level Relation Extraction".

RE_improved_baseline Code for technical report "An Improved Baseline for Sentence-level Relation Extraction". Requirements torch = 1.8.1 transformers

Wenxuan Zhou 74 Nov 29, 2022
Official implementation of our neural-network-based fast diffuse room impulse response generator (FAST-RIR)

This is the official implementation of our neural-network-based fast diffuse room impulse response generator (FAST-RIR) for generating room impulse responses (RIRs) for a given acoustic environment.

12 Jan 13, 2022
A Pytorch Implementation for Compact Bilinear Pooling.

CompactBilinearPooling-Pytorch A Pytorch Implementation for Compact Bilinear Pooling. Adapted from tensorflow_compact_bilinear_pooling Prerequisites I

169 Dec 23, 2022
Out-of-Domain Human Mesh Reconstruction via Dynamic Bilevel Online Adaptation

DynaBOA Code repositoty for the paper: Out-of-Domain Human Mesh Reconstruction via Dynamic Bilevel Online Adaptation Shanyan Guan, Jingwei Xu, Michell

197 Jan 07, 2023
Repository sharing code and the model for the paper "Rescoring Sequence-to-Sequence Models for Text Line Recognition with CTC-Prefixes"

Rescoring Sequence-to-Sequence Models for Text Line Recognition with CTC-Prefixes Setup virtualenv -p python3 venv source venv/bin/activate pip instal

Planet AI GmbH 9 May 20, 2022
Pre-trained Deep Learning models and demos (high quality and extremely fast)

OpenVINO™ Toolkit - Open Model Zoo repository This repository includes optimized deep learning models and a set of demos to expedite development of hi

OpenVINO Toolkit 3.4k Dec 31, 2022
Weakly Supervised Text-to-SQL Parsing through Question Decomposition

Weakly Supervised Text-to-SQL Parsing through Question Decomposition The official repository for the paper "Weakly Supervised Text-to-SQL Parsing thro

14 Dec 19, 2022
A tensorflow=1.13 implementation of Deconvolutional Networks on Graph Data (NeurIPS 2021)

GDN A tensorflow=1.13 implementation of Deconvolutional Networks on Graph Data (NeurIPS 2021) Abstract In this paper, we consider an inverse problem i

4 Sep 13, 2022
Multi-Scale Aligned Distillation for Low-Resolution Detection (CVPR2021)

MSAD Multi-Scale Aligned Distillation for Low-Resolution Detection Lu Qi*, Jason Kuen*, Jiuxiang Gu, Zhe Lin, Yi Wang, Yukang Chen, Yanwei Li, Jiaya J

DV Lab 115 Dec 23, 2022
Simple Tensorflow implementation of "Adaptive Convolutions for Structure-Aware Style Transfer" (CVPR 2021)

AdaConv — Simple TensorFlow Implementation [Paper] : Adaptive Convolutions for Structure-Aware Style Transfer (CVPR 2021) Note This repository does no

Junho Kim 26 Nov 18, 2022
Picasso: A CUDA-based Library for Deep Learning over 3D Meshes

The Picasso Library is intended for complex real-world applications with large-scale surfaces, while it also performs impressively on the small-scale applications over synthetic shape manifolds. We h

97 Dec 01, 2022
Multi-Content GAN for Few-Shot Font Style Transfer at CVPR 2018

MC-GAN in PyTorch This is the implementation of the Multi-Content GAN for Few-Shot Font Style Transfer. The code was written by Samaneh Azadi. If you

Samaneh Azadi 422 Dec 04, 2022
Official code for "On the Frequency Bias of Generative Models", NeurIPS 2021

Frequency Bias of Generative Models Generator Testbed Discriminator Testbed This repository contains official code for the paper On the Frequency Bias

35 Nov 01, 2022
MMDetection3D is an open source object detection toolbox based on PyTorch

MMDetection3D is an open source object detection toolbox based on PyTorch, towards the next-generation platform for general 3D detection. It is a part of the OpenMMLab project developed by MMLab.

OpenMMLab 3.2k Jan 05, 2023
An end-to-end image translation model with weight-map for color constancy

CCUnet An end-to-end image translation model with weight-map for color constancy 1. Download the dataset (take Colorchecker_recommended dataset as an

Jianhui Qiu 1 Dec 21, 2021