Comments (14)
Yes, as pointed before, if you have the requirement to build a large map, implementing loop detection and closure is your choice, otherwise, fusing with other sensors for providing an absolute measurement.
from r3live.
It is hard to say, in our environments, it can close the loops well by itself. The result might show very little difference due to the difference in both software and hardware environments. And, in the sequence of HKUST_campus_00 (1.3 km) and HKUST_campuse_01 (1.5 km), the little drift (e.g., 0.001 deg) at the beginning might cause a large error at the last.
For most of the existing slam systems, no one can guarantee that the result is the same at each time and at each computation platform.
from r3live.
Hi:
Where are the PCD files available to download?
I only found the bag files.
Thanks
from r3live.
Some of our result can be download in this repo: https://github.com/ziv-lin/r3live_preview
from r3live.
Thanks!!
from r3live.
You're right. My notebook has only 32GB of memory, but I noticed that your computer has more than 60 GB of memory. I'll try a better computer later. And I also want to test the upper limit of R3LIVE accuracy by adding the IMU with higher accuracy, such as Xsens 680G. (As the built-in IMU of AVIA is BMI088.)
Again, thanks your great works.
from r3live.
I run the same rosbag in FAST-LIO2, it works fine, there is no obvious height error. (As shown in the following picture.)
I wonder what is the recommended hardware configuration for R3LIVE? Actually, at the last third of HKUST_campus_00 sequence, my physical memory has been used up, can we speculate that the main reason is the lack of physical memory?
from r3live.
It seems that you zoom with different scales, cause the drift in r3live to seem larger? In our testing environment, R3LIVE can stable closes the loop itself.
from r3live.
It is hard to say, is there any loss of message due to the lack of mem?
from r3live.
Is R3LIVE can stable close the loop in other smaller bag?
from r3live.
It seems that you zoom with different scales, cause the drift in r3live to seem larger? In our testing environment, R3LIVE can stable closes the loop itself.
My bad, the history point cloud in ROS_rviz will disappear over time. Here is the correct PCD save by FAST_LIO2. The height error also exists, but still better than other SLAM works.
from r3live.
Both FAST-LIO2, R2LIVE/R3LIVE never expect you can close the loop by traveling up to 1.3 Km in all scenarios at all computation platforms. If you have the requirement to build a large map, implementing loop detection and closure is your choice.
Our paper put this as an example, by saying that we can close the loop itself (at least in our real word experiment, see our recorded video https://www.youtube.com/watch?v=j5fT8NE5fdg) without the aid of additional procedure (i.e., without the aid of loop closure). Aims to show that the drift in our proposed is very low, which demonstrates that our proposed method has good accuracy in localization. This does not contrary to the fact that all slam systems are drifting over time, compare to other slam systems, our proposed system drifts less.
from r3live.
It is hard to say, is there any loss of message due to the lack of mem?
I slow down the play rate of rosbag to 0.5. The delay between rosbag and R3LIVE terminal is gone, but it couldn't close the loop itself. I'll try running it on the server later.
from r3live.
Both FAST-LIO2, R2LIVE/R3LIVE never expect you can close the loop by traveling up to 1.3 Km in all scenarios at all computation platforms. If you have the requirement to build a large map, implementing loop detection and closure is your choice.
Our paper put this as an example, by saying that we can close the loop itself (at least in our real word experiment, see our recorded video https://www.youtube.com/watch?v=j5fT8NE5fdg) without the aid of additional procedure (i.e., without the aid of loop closure). Aims to show that the drift in our proposed is very low, which demonstrates that our proposed method has good accuracy in localization. This does not contrary to the fact that all slam systems are drifting over time, compare to other slam systems, our proposed system drifts less.
I totally agree with you. Compare with other SLAM works, FAST-LIO2 and R3LIVE impress me a lot. I may try to combine the FAST-LIO2 or R3LIVE with global sensors that offer absolute measurements for eliminating drift, such as GPS.
from r3live.
Related Issues (20)
- Similar to issue #82, confusion about point clouds undistortion HOT 2
- 传感器时间的时间标定 HOT 1
- Need help exporting mesh to a simulated environment HOT 5
- .bag数据类型 HOT 2
- livox avia的最短距离如何设置 HOT 1
- Inertial sensors for use with r3live HOT 3
- non-consistant drift between paper and testing run with your uploaded dataset. HOT 1
- Can R3live be reproduced using Livox-mid-70 LiDAR? HOT 1
- 为什么这篇实现里没有使用MTK库而是全部自己实现的卡尔曼滤波呢 HOT 1
- R3LIVE++ HOT 1
- I found unused variable. HOT 1
- Get pointcloud data from ros messages fail HOT 6
- Image input error HOT 2
- Segmentation fault on ikd-tree algorithm HOT 1
- 您好,请问有碰到过这个问题吗? HOT 1
- Can r3live run in the AirSim? HOT 1
- 林博,你好,你有遇到过移动过程中path和odometry抖动得非常剧烈的情况吗?隔一段时间后,产生严重漂移。 HOT 1
- Is it possible to use Livox Mid 360 (a spinning lidar) and two cameras on two side of the lidar and use the algorithm to colorize 360 degrees data? HOT 4
- LIO/VIO cost time HOT 1
- Disable VIO subsystem 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 r3live.