Build bootstrapping for the Motion Compensated Visual-Event Odometry (MC-VEO)
This is the code for the paper [MC-VEO: A Visual-event Odometry with Accurate 6-DoF Motion Compensation].
This source code is GPLv3 license. See the LICENSE file for further details.
- Make sure that the Ruby interpreter is installed on your machine. Rock requires ruby 2.3 or higher, which is provided on Debian and Ubuntu by the ruby2.3 package. This installation process is tested with Ruby 2.7 on Ubuntu 20.04.
docker@hjf $ ruby --version
ruby 2.7.0p0 (2019-12-25 revision 647ee6f091) [x86_64-linux-gnu]
docker@hjf $ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 20.04.4 LTS
Release: 20.04
Codename: focal- Create and “cd” into the directory in which you want to install the toolchain.
docker@hjf $ mkdir mc-veo && cd mc-veo && mkdir dev && cd dev
docker@dev $ pwd
/home/hjf/mc-veo/dev- To build MC-VEO, use this bootstrap.sh script. Save it in the folder you just created.
docker@dev $ wget https://raw.githubusercontent.com/huangfeng95/mc-veo-buildconf/master/bootstrap.sh- In a console run
docker@dev $ sh bootstrap.sh- Follow the installation guide and answer the questions. In case you hesitate choose the answer by default.
Connecting to www.rock-robotics.org (www.rock-robotics.org)|185.199.111.153|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 30078 (29K) [application/octet-stream]
Saving to: 'autoproj_bootstrap'
autoproj_bootstrap 100%[=================================>] 29.37K --.-KB/s in 0.007s
2022-04-13 15:58:06 (3.97 MB/s) - 'autoproj_bootstrap' saved [30078/30078]
Which protocol do you want to use to access huangfeng95/mc-veo-buildconf.git on github.com?
[git|ssh|http] (default: http)
So, what do you want ? (all, none or a comma-separated list of: os gem pip) [all]
Would you like autoproj to keep apt packages up-to-date? [yes]
The current directory is not empty, continue bootstrapping anyway ? [yes]
How should I interact with github.com (git, http, ssh)
If you give one value, it's going to be the method used for all access
If you give multiple values, comma-separated, the first one will be
used for pulling and the second one for pushing. An optional third value
will be used to pull from private repositories (the same than pushing is
used by default) [http,ssh]
Which flavor of Rock do you want to use ?
Stay with the default ('master') if you want to use Rock on the most recent
distributions (Ubuntu 16.04 and later). Use 'stable' only for
now officially unsupported distributions (Ubuntu 14.04) [master]- Answer here 'yes' in case you want to activate python. You need a python installation in your system for this.
Do you want to activate python? [no] yes
Select the path to the python executable [/usr/bin/python3] - Select 'master' in case you want to build MC-VEO with DSO backend or type 'ceres' otherwise
Which flavor of MC-VEO do you want to use ?
Stay with the default ('master') if you want to use MC-VEO with DSO backend
otherwise select ('ceres') [master] - Choose the answers by default in the questions regarding RTT/Orocos. Select 'gnulinux' for Linux based systems
Do you need compatibility with OCL ? (yes or no)
New Rock users that don't need backward compatibility with legacy Orocos components
probably want to say 'no'. Otherwise, say 'yes'.
Saying 'yes' will significantly impact compilation time and the size of the resulting binaries
Please answer 'yes' or 'no' [no]
the target operating system for Orocos/RTT (gnulinux, xenomai, or macosx) [gnulinux] - In case no problems occurred the installation shoudl finish "successfully"
updated environment
Command finished successfully at 2022-04-13 16:47:58 +0200 (took 44 mins 20 secs)- Don't forget to source the environmental variables
docker@dev $ source env.shAll the executing scripts are in bundles/mc-veo directory of your installation.
docker@hjf $ cd mc-veo/dev
docker@dev $ source env.sh
docker@dev $ cd bundles/mc-veoDownload one dataset in log format, for example the atrium dataset:
docker:mc-veo (master) $ wget https://download.ifi.uzh.ch/rpg/eds/atrium/atrium.log -O /tmp/atrium.logRun the run_mc-veo_log.rb script in one terminal
docker@mc-veo (master) $ ruby scripts/slam/mc-veo/run_mc-veo_logs.rb --dataset=atrium --log_type=davis /tmp/atrium.logOpen another terminal and visualize the progress with the mc-veo_visualizaton.rb script.
docker@hjf $ cd mc-veo/dev
docker@dev $ source env.sh
docker@dev $ cd bundles/mc-veo
docker@mc-veo (master) $ ruby scripts/gui/mc-veo_visualization.rb It does not matter the order you run the scripts, visualization or running. The visualization will always wait until the running gives data.
The log files are created in the mc-veo/dev/bundles/mc-veo/tmp folder by default (You can change the folder here). The log files are named in a folder as tmp/date, e.g.: tmp/20220401-1015 MC-VEO is not real time. You can convert the log file to the correct time by running the following command in your log folder
docker@hjf $ cd mc-veo/dev
docker@dev $ source env.sh
docker@dev $ cd bundles/mc-veo/tmp/20220401-1015
docker@20220401-1015 $ rock-convert --use_sample-time mc_veo.0.log The resulting mc_veo.0.log file should be in the newly created updated folder in /tmp/20220401-1015.
You can replay the log files and visualize the results with mc-veo_visualization.rb by doing
docker@hjf $ cd mc-veo/dev
docker@dev $ source env.sh
docker@dev $ cd bundles/mc-veo
docker@mc-veo (master) $ ruby scripts/gui/mc-veo_visualization.rb --log /tmp/20220401-1015/updated/mc_veo.0.log Pocolog files (i.e.: mc-veo.0.log) conversion to Rosbag is possible by running the pocolog2rosbag.py. It is explained in the e2calib repository.
In case you see CORBA related errors when runing the run_mc-veo_log.rb. Restart the omniorb deamon
docker@dev $ sudo /etc/init.d/omniorb4-nameserver stop
docker@dev $ sudo rm -f /var/lib/omniorb/*
docker@dev $ sudo /etc/init.d/omniorb4-nameserver startMC-VEO source code is structure as follows:
- buildconf: is this reposity where the bootstrapping mechanism to install MC-VEO is located.
- bundles/mc-veo: this is the bundles. Bundles are collections of all files needed to run a particular system. Those are configuration files, calibration files and script files for executing the binaries.
- slam/mc-veo: this is the MC-VEO C++ library.
- slam/orogen/mc-veo: this is the Task that builds a class with all the system's functionalities.
- The authors of ESVO
- The authors of EDS
- The authors of Globally_Aligned_Events