DLL: Direct Lidar Localization

Related tags

Deep Learningdll
Overview

DLL: Direct Lidar Localization

Summary

This package presents DLL, a direct map-based localization technique using 3D LIDAR for its application to aerial robots. DLL implements a point cloud to map registration based on non-linear optimization of the distance of the points and the map, thus not requiring features, neither point correspondences. Given an initial pose, the method is able to track the pose of the robot by refining the predicted pose from odometry. The method performs much better than Monte-Carlo localization methods and achieves comparable precision to other optimization-based approaches but running one order of magnitude faster. The method is also robust under odometric errors.

DLL is fully integarted in Robot Operating System (ROS). It follows the general localization apparoch of ROS, DLL makes use of sensor data to compute the transform that better fits the robot odometry TF into the map. Although an odometry system is recommended for fast and accurate localization, DLL also performs well without odometry information if the robot moves smoothly.

DLL experimental results in different setups

Software dependencies

There are not hard dependencies except for Google Ceres Solver and ROS:

Hardware requirements

DLL has been tested in a 10th generation Intel i7 processor, with 16GB of RAM. No graphics card is needed. The optimization is currently configured to be single threaded. You can easily reduce the processing time by a 33% just increasing the number of threads used by Ceres Solver.

Compilation

Download this source code into the src folder of your catkin worksapce:

$ cd catkin_ws/src
$ git clone https://github.com/robotics-upo/dll

Compile the project:

$ cd catkin_ws
$ source devel/setup.bash
$ catkin_make

How to use DLL

You can find several examples into the launch directory. The module needs the following input information:

  • A map of the environment. This map is provided as a .bt file
  • You need to provide an initial position of the robot into the map.
  • base_link to odom TF. If the sensor is not in base_link frame, the corresponding TF from sensor to base_link must be provided.
  • 3D point cloud from the sensor. This information can be provided by a 3D LIDAR or 3D camera.
  • IMU information is used to get roll and pitch angles. If you don't have IMU, DLL will take the roll and pitch estimations from odometry as the truth values.

Once launched, DLL will publish a TF between map and odom that alligns the sensor point cloud to the map.

When a new map is provided, DLL will compute the Distance Field grid. This file will be automatically generated on startup if it does not exist. Once generated, it is stored in the same path of the .bt map, so that it is not needed to be computed in future executions.

As example, you can download 5 datasets from the Service Robotics Laboratory repository (https://robotics.upo.es/datasets/dll/). The example launch files are prepared and configured to work with these bags. You can see the different parameters of the method. Notice that, except for mbzirc.bag, these bags do not include odometry estimation. For this reason, as an easy work around, the lauch files publish a fake odometry that is the identity matrix. DLL is faster and more accurate when a good odometry is available.

Cite

DLL has been accepted for publication in IROS 2021.

F. Caballero and L. Merino. "DLL: Direct LIDAR Localization. A map-based localization approach for aerial robots". Sumbitted to the International Conference on Intelligent Robots and Systems, IROS 2021.

You can download preliminar version of the the paper from arXiv

Comments
  • Using Livox mid 70 get bad result

    Using Livox mid 70 get bad result

    Hi, I use Livox mid 70 with wheel odometry and IMU, but the localization result is not good, the robot pose always "jump" when running. any idea to make a better result (stable, smooth, continues path)

    opened by gongyue666 9
  • Run other datasets

    Run other datasets

    hello!I saved a .ot file in dll/maps. And <arg name="map" default="myown.ot" /> But when I run the program , it shows "NULL otcomap". How come?Where else do I need to set the path?

    opened by MIke-1118 6
  • tested the given bag failed

    tested the given bag failed

    Hi, thanks for your great work! I have download the given bag for test the dll,but when i launched the launch file,it always shows the error,which is : " Octomap loaded Map size: x: 37.2 to 92.75 y: 41.95 to 95.65 z: -10.4 to 0.15 Res: 0.05 Error opening file /home/whx/study/dll_ws/src/dll/maps/airsim.grid for reading Computing 3D occupancy grid. This will take some time... [ INFO] [1640669470.668451692, 1614448809.604375476]: Progress: 0.000000 % [ INFO] [1640669471.163893210, 1614448810.107720910]: Progress: 0.021567 % [ INFO] [1640669471.668560708, 1614448810.612384198]: Progress: 0.039648 % [ INFO] [1640669472.172075265, 1614448811.115887848]: Progress: 0.053874 % [ INFO] [1640669472.680451449, 1614448811.624293216]: Progress: 0.065055 % [ INFO] [1640669473.184041975, 1614448812.127884273]: Progress: 0.073926 % ... ... [bag_player-2] process has finished cleanly log file: /home/whx/.ros/log/5879e12a-679f-11ec-9f57-c0e43482dfff/bag_player-2*.log " I have noticed there is a closed issue which talk about it,so i repeated the same test for many times.But it didn't work.

    I hope someone can help me solve the problem.

    Best wishes

    opened by numb0824 2
  • open map file failed

    open map file failed

    Thanks for your great works! I want to run your code just used roslaunch dll airsim1.launch and changed the true path about the .bag. But I meet the following error Screenshot from 2021-11-30 10-16-11 Could you help me how to solve the problem? Thanks.

    opened by huangsiyuan0717 2
  • Transform of input map

    Transform of input map

    Hello!

    I'd first like to thank you for this work, it's very interesting!

    I have a question regarding the internal representation of the map: when looking through the code I notice that you subtract the minimum values from each axis of the points. I suppose this is relevant for the method? I got some (obviously) poor results when I assumed the input map and internal representation were the same.

    I think it would be nice to make this clearer in the readme, or potentially add some transform between the original map and the internal representation such that the initial position set in the launch file could be relative the original map.

    opened by MartinEekGerhardsen 3
Releases(v1.1)
  • v1.1(Mar 22, 2022)

    Improved memory allocation and solver parameterization

    • Added use_yaw_increments parameter that uses yaw increments from IMU since last LIDAR update as initial guess for the optimizer. This is a good choice when robot performs very fast yaw rotations
    • Added grid trilinear interpolation computation online. This will reduce the DLL memory requirements by a factor of 7 approximatelly
    • Added parameters to set solver max iterations and max threads
    • Added comprehensive message when .grid files is no found
    Source code(tar.gz)
    Source code(zip)
  • v1.0(Mar 22, 2022)

    Initial Commit

    • This version contains the source code related wit the IROS paper detailed in the README
    • Some cleaning has been done to make it simpler to understand
    Source code(tar.gz)
    Source code(zip)
Owner
Service Robotics Lab
Service Robotics, Autonomous Robot Navigation, Machine Learning, Social Robotics
Service Robotics Lab
Semi-supervised semantic segmentation needs strong, varied perturbations

Semi-supervised semantic segmentation using CutMix and Colour Augmentation Implementations of our papers: Semi-supervised semantic segmentation needs

146 Dec 20, 2022
CTF challenges and write-ups for MicroCTF 2021.

MicroCTF 2021 Qualifications About This repository contains CTF challenges and official write-ups for MicroCTF 2021 Qualifications. License Distribute

Shellmates 12 Dec 27, 2022
Syllabus del curso IIC2115 - Programación como Herramienta para la Ingeniería 2022/I

IIC2115 - Programación como Herramienta para la Ingeniería Videos y tutoriales Tutorial CMD Tutorial Instalación Python y Jupyter Tutorial de git-GitH

21 Nov 09, 2022
Semantic segmentation models, datasets and losses implemented in PyTorch.

Semantic Segmentation in PyTorch Semantic Segmentation in PyTorch Requirements Main Features Models Datasets Losses Learning rate schedulers Data augm

Yassine 1.3k Jan 07, 2023
RNG-KBQA: Generation Augmented Iterative Ranking for Knowledge Base Question Answering

RNG-KBQA: Generation Augmented Iterative Ranking for Knowledge Base Question Answering Authors: Xi Ye, Semih Yavuz, Kazuma Hashimoto, Yingbo Zhou and

Salesforce 72 Dec 05, 2022
Norm-based Analysis of Transformer

Norm-based Analysis of Transformer Implementations for 2 papers introducing to analyze Transformers using vector norms: Kobayashi+'20 Attention is Not

Goro Kobayashi 52 Dec 05, 2022
Repository aimed at compiling code, papers, demos etc.. related to my PhD on 3D vision and machine learning for fruit detection and shape estimation at the university of Lincoln

PhD_3DPerception Repository aimed at compiling code, papers, demos etc.. related to my PhD on 3D vision and machine learning for fruit detection and s

lelouedec 2 Oct 06, 2022
Teaches a student network from the knowledge obtained via training of a larger teacher network

Distilling-the-knowledge-in-neural-network Teaches a student network from the knowledge obtained via training of a larger teacher network This is an i

Abhishek Sinha 146 Dec 11, 2022
Python implementation of "Single Image Haze Removal Using Dark Channel Prior"

##Dependencies pillow(~2.6.0) Numpy(~1.9.0) If the scripts throw AttributeError: __float__, make sure your pillow has jpeg support e.g. try: $ sudo ap

Joyee Cheung 73 Dec 20, 2022
Codebase for Time-series Generative Adversarial Networks (TimeGAN)

Codebase for Time-series Generative Adversarial Networks (TimeGAN)

Jinsung Yoon 532 Dec 31, 2022
Spatial Single-Cell Analysis Toolkit

Single-Cell Image Analysis Package Scimap is a scalable toolkit for analyzing spatial molecular data. The underlying framework is generalizable to spa

Laboratory of Systems Pharmacology @ Harvard 30 Nov 08, 2022
Automatic Video Captioning Evaluation Metric --- EMScore

Automatic Video Captioning Evaluation Metric --- EMScore Overview For an illustration, EMScore can be computed as: Installation modify the encode_text

Yaya Shi 17 Nov 28, 2022
一个多语言支持、易使用的 OCR 项目。An easy-to-use OCR project with multilingual support.

AgentOCR 简介 AgentOCR 是一个基于 PaddleOCR 和 ONNXRuntime 项目开发的一个使用简单、调用方便的 OCR 项目 本项目目前包含 Python Package 【AgentOCR】 和 OCR 标注软件 【AgentOCRLabeling】 使用指南 Pytho

AgentMaker 98 Nov 10, 2022
A diff tool for language models

LMdiff Qualitative comparison of large language models. Demo & Paper: http://lmdiff.net LMdiff is a MIT-IBM Watson AI Lab collaboration between: Hendr

Hendrik Strobelt 27 Dec 29, 2022
Joint learning of images and text via maximization of mutual information

mutual_info_img_txt Joint learning of images and text via maximization of mutual information. This repository incorporates the algorithms presented in

Ruizhi Liao 10 Dec 22, 2022
Prompt-BERT: Prompt makes BERT Better at Sentence Embeddings

Prompt-BERT: Prompt makes BERT Better at Sentence Embeddings Results on STS Tasks Model STS12 STS13 STS14 STS15 STS16 STSb SICK-R Avg. unsup-prompt-be

196 Jan 08, 2023
Attention for PyTorch with Linear Memory Footprint

Attention for PyTorch with Linear Memory Footprint Unofficially implements https://arxiv.org/abs/2112.05682 to get Linear Memory Cost on Attention (+

11 Jan 09, 2022
Semantic Segmentation of images using PixelLib with help of Pascalvoc dataset trained with Deeplabv3+ framework.

CARscan- Approach 1 - Segmentation of images by detecting contours. It failed because in images with elements along with cars were also getting detect

Padmanabha Banerjee 5 Jul 29, 2021
DatasetGAN: Efficient Labeled Data Factory with Minimal Human Effort

DatasetGAN This is the official code and data release for: DatasetGAN: Efficient Labeled Data Factory with Minimal Human Effort Yuxuan Zhang*, Huan Li

302 Jan 05, 2023
The code for Expectation-Maximization Attention Networks for Semantic Segmentation (ICCV'2019 Oral)

EMANet News The bug in loading the pretrained model is now fixed. I have updated the .pth. To use it, download it again. EMANet-101 gets 80.99 on the

Xia Li 李夏 663 Nov 30, 2022