M2SLAM

Authors: Fu Li, Shaowu Yang

Current version: 1.0.0

introduction

M2SLAM is a novel visual SLAM system with memory management to overcome two major challenges in reducing memory con- sumption of visual SLAM: efficient map data scheduling between the memory and the external storage, and map data persistence method (i.e., the data out- lives the process that created it). We redesign the framework of a visual SLAM system to contain a SLAM front end and a database server end. The front end maintains a localization and a mapping process, as well as map data scheduling between the memory and the database. We propose a map data scheduling method to organize map data considering the local spatial character among keyframes. The server end provides data services from a spatial database for global map data storage and accessing.

The framework of M2SLAM is shown above, which includes the front end and the server end. We implement the front end based on ORB_SLAM2[1-2] by extending its tracking, local mapping, and loop closing module, and adding one novel memory managing module. The server end publishes data services for map data storage and accessing.

0. Related Publications

1. Prerequisites

1.0 requirements

ubuntu 14.04
ROS indigo
ORBSLAM2 1.0.0
boost
ODB 2.4.0 or later
PostgreSQL 9.4 or later
PostGIS 2.2 or later

1.1 ROS install

Inatall ROS indigo according to the instructions in ROS wiki.

1.2 ORBSLAM2 and its dependencies

Our M2SLAM system is build on the foundation of [ORB_SLAM2(https://github.com/raulmur/ORB_SLAM2). You should follow the instructions provided by ORB_SLAM2 build its dependencies. We do not list here.

1.3 Boost library install

We use boost library to serialize and deserialize the data. We can install boost library using the following instruction in terminal.

sudo apt-get instal libboost-dev

1.4 ODB environment build

We use ODB to solve the issue of mapping a C++ object to a record in a database table, instead of using SQL.

You can use the ODB packages in src/orbslam_client/Thirdparty/package or download from the ODB official website.

Unpack the packages and build the ODB according to the INSTALL file in their dirs.

The following errors may occur:

configure : error: g++ does not support plugins; reconfigure GCC with --enable-plugin

sudo apt-get update
sudo apt-get install gcc-4.8-plugin-dev

error: libcutl is not found

Download the source code of the libcutl from its office website.

Install the package according to the INSTALL file.

Can not find odb, when test the odb-example

The reason is that the install directory /usr/local/lib has not in the ld.so.conf file.

echo "/usr/local/lib" >> /etc/ld.so.conf
echo "/usr/local/bin" >> /etc/ld.so.conf
sudo ldconfig

1.5 PostGIS install

sudo apt-get install postgresql-9.4
sudo apt-get install postgresql-9.4-postgis-2.2

2. Building and Run M2SLAM

2.1 create PostGIS environment

In PostGIS, we should create the USER and DATABASE using PGAdmin or terminal. We create the USER:m2slam_usr and DATABASE:m2slam_db.

2.2 create odb environment and create tables in the DATABASE m2slam_db

Use odb to compile the basic data model in orbslam_servcer

Terminal in src/orbslam_servcer/include directory:

odb -d pgsql --generate-query --generate-schema Data_TopoKeyFrame.h
odb -d pgsql --generate-query --generate-schema Data_TopoMapPoint.h
odb -d pgsql --generate-query --generate-schema Data_KeyFrame.h
odb -d pgsql --generate-query --generate-schema Data_MapPoint.h
psql -U m2slam_usr -h 127.0.0.1 -d m2slam_db < Data_TopoKeyFrame.sql
psql -U m2slam_usr -h 127.0.0.1 -d m2slam_db < Data_TopoMapPoint.sql
psql -U m2slam_usr -h 127.0.0.1 -d m2slam_db < Data_KeyFrame.sql
psql -U m2slam_usr -h 127.0.0.1 -d m2slam_db < Data_MapPoint.sql

You can verify that the tables have been created in the database m2slam_db.

2.3 build M2SLAM

The M2SLAM runs as the ROS package. and the M2SLAM src directory should be the ROS package directory, catkin_src/. We use catkin tool to organize the M2SLAM packages, orbslam_client and orbslam_servcer.

Terminal in the catkin_src/ directory.

catkin_make
catkin_install

2.4 run M2SLAM

2.4.1 start ros core

ros core

2.4.2 start orbslam_servcer

rosrun orbslam_servcer orbslam_servcer

2.4.3 run orbslam_client in different datasets

run TUM RGB-D datasets

Download a sequence from http://vision.in.tum.de/data/datasets/rgbd-dataset/download and uncompress it.

Associate RGB images and depth images using the python script associate.py. We already provide associations for some of the sequences in Examples/RGB-D/associations/. You can generate your own associations file executing:

python associate.py PATH_TO_SEQUENCE/rgb.txt PATH_TO_SEQUENCE/depth.txt > associations.txt

Execute the following command. Change TUMX.yaml to TUM1.yaml,TUM2.yaml or TUM3.yaml for freiburg1, freiburg2 and freiburg3 sequences respectively. Change PATH_TO_SEQUENCE_FOLDERto the uncompressed sequence folder. Change ASSOCIATIONS_FILE to the path to the corresponding associations file.

rosrun ORB_SLAM2_client_TUM_rgbd ORB_SLAM2_client_TUM_rgbd PATH_TO_VOCABULARY PATH_TO_SETTINGS_FILE ONLINE_RECTIFICATION

run KITTI datasets

Download the dataset (grayscale images) from http://www.cvlibs.net/datasets/kitti/eval_odometry.php

Execute the following command. Change KITTIX.yamlto KITTI00-02.yaml, KITTI03.yaml or KITTI04-12.yaml for sequence 0 to 2, 3, and 4 to 12 respectively. Change PATH_TO_DATASET_FOLDER to the uncompressed dataset folder. Change SEQUENCE_NUMBER to 00, 01, 02,.., 11.

rosrun ORB_SLAM2_client_KITTI_stereo ORB_SLAM2_client_KITTI_stereo Vocabulary/ORBvoc.txt Examples/Stereo/KITTIX.yaml PATH_TO_DATASET_FOLDER/dataset/sequences/SEQUENCE_NUMBER

Reference

[1] Mur-Artal R, Montiel J M M, Tardos J D. ORB-SLAM: a versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5): 1147-1163.

[2] Mur-Artal R, Tardos J D. ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras[J]. arXiv preprint arXiv:1610.06475, 2016.

License

M2SLAM is released under a GPLv3 license.

For a closed-source version of M2SLAM for commercial purposes, please contact the authors.

lifunudt / M2SLAM