Comments (32)
Hi!
Thanks for your interest in DOOR-SLAM!
Please make sure to use the git tag DOOR-SLAM-1.0
when building the system. https://github.com/lajoiepy/Buzz_SLAM/tree/DOOR-SLAM-1.0
The new version is not compatible.
I will shortly fix the Dockerfiles accordingly!
from door-slam.
Thanks for your reply!!!
if i use DOOR-SlAM-1.0,i got the problem below.
And i found this library in another author's repository .But when i tried to catkin_make in the file of ros_ws from this author's repository,i got the new error below.
i have no idea how to slove it.
Thanks so much for your reply!!!
from door-slam.
To build the whole system, please use this Dockerfile and instructions instead : https://github.com/bramtoula/multi_robot_SLAM_separators/tree/master/docker/camera_and_slam_tx2
from door-slam.
Thank you so much for your patience!!!!
Should I use the Dockerfile you mentioned if I just want to get the result of simulation only with my computer.Because I found the Dockerfile starts with "lajoiepy/realsense-tx2-px4".
Looking forward to your reply!!!
from door-slam.
For simulation only, please use this Dockerfile and instructions
from door-slam.
Sorry to waste you so mush time.
I may not be able to express what I mean in vain.
All errors comes from the Dockerfile that is for simulation only .
When I tried to simulate,I got these errors.They are all from here.
If i use DOOR-SLAM-1.0,I got an error mentioned before when I try to install Buzz SLAM .
Thank you so much for your reply!!
from door-slam.
No problem :) Can you try the docker build with the --no-cache
option? do you still have the same issue?
from door-slam.
yes.I still get the same error.
from door-slam.
When I docker build it again.I got the same problem.
At the same time,it shows "No package argos_simulator found".I have no idea whether this has any effect on the result.
Looking forward to you reply!!!
from door-slam.
The simulator and third party libraries evolve continuously and DOOR-SLAM is no longer updated. For a stable experience to reproduce our results, I strongly suggest you to use the working docker image lajoiepy/doorslam-argos-simulation
that we provide. If you are unfamiliar with Docker you can run docker pull lajoiepy/doorslam-argos-simulation
to download it.
from door-slam.
Thank you so much!!!The problem has been solved.
I want to test kitti on TX2 now.Could you please tell me how can I get the result? I have already pulled images on TX2.
Looking forward to your replay!!!
from door-slam.
Hi!
First follow the setup here: https://github.com/MISTLab/DOOR-SLAM/blob/master/tx2-setup/docker/instructions.md
Then, split your KITTI ROS bag into 3 parts named bag-0.bag
, bag-1.bag
, and bag-2.bag
. Put one on each TX2 (you need 3 on the same network). Then launch the kitti_params
commands from the aliases here.
Have a nice day!
from door-slam.
Thank you for your reply!!!!
I still have some problems.
If I have 2 TX2,is it necessary to split KITTI into 3 parts? And I have no idea where I should save these dataset exactly.(on TX2 itself or in a container?)
About setting paremeters, are they related to network like MASK、OWN_IP .
When I try to launch kitti_params,the terminal says that the container is not running.And when I run images you provided ,I find the container exits immediately.And I run images like sudo docker run -it lajoiepy/doorslam-realsensed435-tx2-dji.
Did I do something wong?
Have a good day!!!
from door-slam.
You can do it on 2 TX2, look closely the bash commands in .bash_aliases to understand how it works (where to place files, etc.)
Before launch kitti_params
you need a container running, to do this launch docker_bash
first.
from door-slam.
Thank you so much!
When I test kitti ,can I get other results except pose graph?like point cloud map?
from door-slam.
No, DOOR-SLAM doesn't currently support the production of a 3D map. However, you could save the 3D pointclouds from your sensors and use the pose graph to align them in post-processing.
from door-slam.
I got it!!!
If I want to download KITTI dataset,I find the website according to A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? the KITTI vision benchmark suite,” in IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Providence, USA, June 2012, pp. 3354–3361.
But I have no idea which dataset I should download?Is this right place to download dataset you mentioned in your paper?(http://www.cvlibs.net/datasets/kitti/eval_stereo_flow.php?benchmark=stereo)
Thank you a lot!!!
from door-slam.
It is here: http://www.cvlibs.net/datasets/kitti/eval_odometry.php :)
from door-slam.
I got it!
Thank you so much.
I try to download grayscale one.And I find it has left images and right images.now I can convert just one of them into .bag.I have no idea how to covert all of them into one .bag.
And if I split BAG into two parts,how much overlaps shouldthey have?
thank you for your help!
from door-slam.
Hi!
Sorry to take up your time.I am eager for the results.
I successfully put a Kitti Ros bag on one Tx2.But when I run Kitti-param ,I got some errors like "error while loading share libraries:libpcl.so.1.8:cannot open shared object file:No such file or directory"and "Import Error:libcublas.so.10.0:cannot open shared object file:No such file or directory"
I flash the Tx2 with jetpack4.6.I guess this may not lead to these errors.
Looking for forward to your reply !
from door-slam.
We were using Jetpack 4.2 for the experiments, this might be the issue here.
from door-slam.
THANK YOU SO MUCH!
The problem has been solved.
I download a dataset from internet and this is the information about this .bag.(this bag's topic is a little bit different from what we want.for example,/kitti/camera_color_left/image_raw.What we want is /kitti/camera_color_left/image_rect.So I change the name of topic to what we want with the help of "rosbag record ")
And when I run kitti_param,I get an error below.
It seems like DOOR-SLAM can not recieve data from this topic.But when I use rostopic hz my_topic,these topics are sending messages.
I use this dataset to test ,because I don not kown how to transfer left images and right images into one rosbag.
Thank you for your reply!!!
from door-slam.
The problem has been solved!
it seems that I get some results.
could you tell me how I can get a pose graph pictures? I want to use rviz,but docker can not connect to X display.
from door-slam.
The resulting pose graph files should be in the datasets
folder you created. They are in the g2o format (https://github.com/RainerKuemmerle/g2o). We do not provide a visualization tool, but you could use the one from g2o or others or make your own .
from door-slam.
I got it!THANK YOU!!!
I want to add intel realsense to this experiment.But I use AGX Xavier.When I run ./buildPatchedKernel.sh,it does not provide L4Target 32.1.0.
Is this enough if I change
“if [ "$JETSON_BOARD" == "AGX Xavier" ] ; then
L4TTarget="31.1.0"
Test for 31.1.0 first
if [ $JETSON_L4T = "31.1.0" ] ; then
KERNEL_BUILD_VERSION=vL4T31.1.0”
to
if [ "$JETSON_BOARD" == "AGX Xavier" ] ; then
L4TTarget="32.1.0"
Test for 31.1.0 first
if [ $JETSON_L4T = "32.1.0" ] ; then
KERNEL_BUILD_VERSION=vL4T32.1.0
(ps:31.1.0 to 32.1.0)
Thank you for your reply!!!
from door-slam.
hi! I find if I use jetpack4.2 to flash the agx xavier.I got the L4T 32.1 which is not included in your buildpached kernel.sh.But I found you take agx
with L4T 31.1 into consider.
I Have to flash agx with jetpack 4.1.1 developer preview if I want to get L4T 32.1. but this is not the same with your suggestion about jetpack4.2.
I have no idea what going on here?Looking forward to your reply!thank you so much!
from door-slam.
We were using the NVIDIA Jetson TX2 instead of the Xavier AGX, hence the difference in JetPack version. Unfortunately, I can not provide support on the Xavier AGX setup.
from door-slam.
THANK YOU FOR YOUR REPLY!
I notice that in your readme.md you said it is for librealsense v2.22.0.
But I see LIBREALSENSE_VERSION=v2.24.0 in your installLibrealsense.sh.
so the version of librealsense I got last is v2.24.0 .Am I right?
Is batman-adv module necessery?Because it is a little hard to install this on agx.
Thank you sooooo much!
from door-slam.
I don't think the exact version of the librealsense is important. You should take the latest working version for your computer and camera. batman-adv
is not needed if you use a centralized network like wifi.
from door-slam.
OK!THANK YOU!
I have two questions about this experiment.
Will I still get only pose graph result in datasets that I created If I use camera on TX2.If I want to visualize result,I also need to do it by myself.
The second question is that when I test kitti on TX2 last time, I find arithmetic speed is very slow.Is this normal?or right? How is the real-time nature of this project?
THANK YOU!!!
from door-slam.
The loop closures should be detected in real-time (or close to) if you have perfect communication. However if there are significant delays on your network, this makes the process slower. The pose graph optimization takes more time (usually less than a minute depending on the stopping threshold and the communication delays) and should be done periodically similar to other SLAM systems.
from door-slam.
THANKS!!!
I tried to connect camera on agx,and I run the command "camera".But the docker told me that it can not find the divice.I notice that the version of realsense ros is v2.27.0.But I find this version only support ubuntu 16.04.However in your docker image,it is ubuntu 18.04.I have no idea whether this may cause the error.
Thank you so much for your help!!!
from door-slam.
Related Issues (13)
- Is that possible to test it with MacOS and Docker? HOT 3
- Question about launch command 'blabbermouth2' HOT 5
- can't successfully build the package HOT 3
- Why is RANSAC works well in centralized setting? HOT 1
- pcm src HOT 2
- QXcbConnection: XCB error: 13 HOT 6
- How could I set up the 3 robot SLAM? HOT 2
- here is SLAM framework
- where is the SLAM framework HOT 2
- Question about testing on KITTI HOT 1
- Question in launching container HOT 2
- build bug HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from door-slam.