srv / stereo_slam Goto Github PK
View Code? Open in Web Editor NEWLicense: Other
License: Other
I am trying to use a stereo slam package,and by following the instructions, I completed the installation. I am trying to run this slam with drcsim Atlas bot, spawned in a gazebo vrc task environment.
Now, by editing the demo.launch file of the package, I can link the stereo_slam package to subscribe to topics which provide with the stereo pair of images. The edited launch file
<launch>
<arg name="bagfile" default="/tmp/ROS.bag"/>
<!-- Setup the camera -->
<arg name="camera" default="/multisense_sl/camera/"/>
<param name="/use_sim_time" value="true"/>
<!-- Play the bagfile -->
<node pkg="rosbag" type="play" name="rosbag" args="--clock $(arg bagfile)"/>
<!-- Run the stereo image proc -->
<group ns="$(arg camera)">
<node pkg="stereo_image_proc" type="stereo_image_proc" name="stereo_image_proc" />
</group>
<!-- Run the visual odometry: You can choose between viso2 and fovis -->
<node pkg="viso2_ros" type="stereo_odometer" name="stereo_odometer">
<remap from="stereo" to="$(arg camera)"/>
<remap from="image" to="image_color"/>
<param name="base_link_frame_id" value="/pelvis"/>
</node>
<!--node pkg="fovis_ros" type="stereo_odometer" name="fovis">
<remap from="stereo" to="$(arg camera)"/>
<remap from="image" to="image_color"/>
<param name="base_link_frame_id" value="/pelvis"/>
</node-->
<!-- Stereo SLAM -->
<node pkg="stereo_slam" type="localization" name="slam" output="screen">
<!-- Working directory -->
<param name="work_dir" value="/home/tonystark/freelancing/Atlas_slam/slam/" />
<!-- Topic parameters -->
<param name="odom_topic" value="stereo_odometer/odometry"/>
<param name="left_topic" value="$(arg camera)/left/image_color"/>
<param name="right_topic" value="$(arg camera)/right/image_color"/>
<param name="left_info_topic" value="$(arg camera)/left/camera_info"/>
<param name="right_info_topic" value="$(arg camera)/right/camera_info"/>
<!-- Min displacement between graph vertices (in meters) -->
<param name="min_displacement" value="0.25"/>
<!-- Descriptor type: SIFT or SURF -->
<param name="desc_type" value="SIFT"/>
<!-- Can be "CROSSCHECK" or "RATIO" -->
<param name="desc_matching_type" value="CROSSCHECK"/>
<!-- Descriptor threshold for crosscheck matching (typically between 0.7-0.9) or ratio for ratio matching (typically between 0.6-0.8). -->
<param name="desc_thresh_ratio" value="0.88" />
<!-- Minimum number of neighbors that will be skiped for the loop closure (tipically between 5-20, but depends on the frame rate) -->
<param name="min_neighbor" value="10" />
<!-- Get the n first candidates of the hash matching (tipically between 1-5) -->
<param name="n_candidates" value="5" />
<!-- Minimun number of descriptor matches to consider a matching as possible loop closure (>8) -->
<param name="min_matches" value="30" />
<!-- Minimum number of inliers to consider a matching as possible loop closure (>8) -->
<param name="min_inliers" value="15" />
<!-- Set to 0 for LinearSlam Solver with gauss-newton. Set to 1 for LinearSlam Solver with Levenberg -->
<param name="g2o_algorithm" value="0" />
<!-- Maximum number of iteration for the g2o algorithm -->
<param name="g2o_opt_max_iter" value="10" />
</node>
<!-- Pointcloud collector -->
<node pkg="stereo_slam" type="collector" name="collector" output="screen">
<param name="work_dir" value="/home/tonystark/freelancing/Atlas_slam/slam/" />
<param name="graph_topic" value="slam/graph"/>
<param name="cloud_topic" value="$(arg camera)/scaled/points2"/>
<rosparam file="$(find stereo_slam)/etc/reconstruction.yaml" />
</node>
When executing
roslaunch stereo_slam demo.launch
I get the following error
error loading <rosparam> tag:
file does not exist [/home/tonystark/lib_haloc/src/stereo_slam/etc/reconstruction.yaml]
XML is <rosparam file="$(find stereo_slam)/etc/reconstruction.yaml"/>
The traceback for the exception was written to the log file
Upon searching, neither directory /etc nor the file reconstruction.yaml exited. Has anybody made use of this package? If yes, can you please tell me how to solve this? Is it a file and folder I should create by myself? Thanks in advance
Hey guys,
If one uses the corrected odometry output for vehicle control, how would the pose output behave on loop closures? I guess there would be a jump if there graph correction is done starting from the previous nodes ?
Or is the correction applied from the front nodes, thus only affecting the older graph points and there is no odometry jump?
Kabir
Would be possible to upload a complete and working Indigo packet? I've tried doing a lot of changes and correcting one by one the errors that I found while compiling, but it is not possible to fix it at all.
My brain is goint to explode.
Thank you in advance.
This is a great package, thanks for publishing your code! I'm having a little bit of trouble with camera calibration though. Right now I am calibrating the camera in the water (onboard the ROV) using the ROS camera calibration system. It's still providing very poor results, do you have a recommended method for properly calibrating a stereo array to utilize this package? Thanks again.
I tried to build stereo SLAM from kinetic branch then I got the following error
error: ‘class haloc::Hash’ has no member named ‘isInitialized’
if (!hash_.isInitialized()
It seems that the function is belong to libhaloc indigo version not kinetic version
Project 'stereo_slam' tried to find library 'haloc'. The library is
neither a target nor built/installed properly. Did you compile project
'libhaloc'? Did you find_package() it before the subdirectory containing
its code is included?
Call Stack (most recent call first):
CMakeLists.txt:33 (find_package)
Hello i am trying to catkin_make your package. However i get errors, due to opencv dependencies. Which version of opencv do i need?
Thank you
Hello! Thanks for the code.
Has the collector
node that is launched by the demo been deprecated?
If so are there any pointcloud collector/ publishers that you might recommend?
would be great to indicate in the doc which device is compatible ?
do I build my own rig with 2 webcams or do I need to purchase a kinect or something ?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.