Comments (3)
@henn1001 several points that you should take a close look at:
- you do need to provide the IMU to camera transformation. What the framework expects here is the position and orientation of the camera w.r.t. the IMU frame of reference, expressed in the IMU frame of reference. This is the first and most important part to fix.
absolute measurements
refers to the fact that your measurements are not a relative quantity but denote the position of the camera system wrt. some global base-frame. You will need to switch this parameters totrue
- don't fix the scale factor. Set
pose_sensor/pose/fixed_scale
tofalse
(also you have this setting twice in your yaml, so you need to clean these up too) pose_sensor/pose_measurement_world_sensor
needs to be double checked. If LSD slam provides the position/orientation of the camera wrt. a "world" frame of reference then this must be set to true. If LSD slam provides the position/orientation of a "world" frame expressed in the camera frame of reference, then this must be set to false.- Yes, if the covariance of LSD is not set, then you should use the
fixed-covariance
setting in the estimator. - The dropouts on the IMU you are seeing are probably due to your PC being too slow or your IMU node having in fact dropouts. If this is just about single messages, then it should work still fine.
from ethzasl_msf.
Thanks for your reply and your help - really appreciate that
- Yeah i understand this part. Since my Imu is sitting directly under my camera and the frames are already aligned, i wont have to rotate anything right? so i should be good with 1,0,0,0 / w,x,y,z ? i'm not that good with quaternions : )
- OK so if i set my position/vslam sensor as a absolute measurement - the thing that will be wrong is the scale. that's why i shouldn't use a fixed scale and where "pose_initial_scale:" comes into play. But over time the scale should automatically converge to a metric scale right?
- "pose_measurement_world_sensor: false" should be correct - same as ptam
- i fixed the imu dropouts -> had to set up a queue in my publisher node
Also i did some testing:
This is the position output from lsd_slam. after the initialization i moved it back and forth ~40 cm (x axis) and then to the side ~60 cm (y axis)
MSF output with absolute measurements = true
and here with a initial_scale = 0.5
from ethzasl_msf.
Hi, @henn1001 Have you managed to get your implementation of msf setup to work?
from ethzasl_msf.
Related Issues (20)
- Set up with LSD SLAM and front-looking camera on quadcopter in Gazebo
- msf State Initialization HOT 1
- What's exactly the definition of q?
- ERROR: cannot launch node of type [msf_updates/pose_sensor]: can't locate node [pose_sensor] in package [msf_updates]
- Issues about sensor_fusion
- Template mismatch under vs 2017 HOT 1
- Drift in pose estimation HOT 1
- Simple Guide for windows VS 2017 build
- IMU is working wrong? Parameters?
- tf broadcaster
- GPS + SLAM, P quickly diverges.
- Raw imu input witjout orientation
- Graph not appearing HOT 1
- No such file or directory
- Is it possible to Fuse IMU and Pressure (barometer) data? HOT 1
- Fusion IMU+LIDAR+CAMERA
- Tutorial graph not appearing
- Can the filtered trajectory be saved?
- 10-year anniversary of SSF/MSF HOT 2
- Not able to reset password 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 ethzasl_msf.