Closed wmlynar closed 5 years ago
Hello Wojciech and Merry Xmas!
Thanks for comments.
This repo is a test repo right now, but I am refactoring the code in order to properly release it and show how to fuse kinematically computed odometry (e.g. leg odometry for humanoids, wheel odometry for mobile robots), visual odometry and/or any other pose i.e. GPS readings with the IMU readings.
Hopefully, this will be done in about 1-2 months or so, considering we have Christmas holidays and this project is on-going research! On the other hand I'let you know that in my lab tests so far I was able to accurately estimate the 3D-pose, 3D-velocity and IMU biases for my walking humanoid robot so I am pretty confident that this filter works when properly tuned!
The EKF is based on https://groups.csail.mit.edu/rrg/papers/icra12_aggressive_flight.pdf but with the imu biases in the state vector.
I'll leave this issue open so I can get back to you when I have the latest version. In the meanwhile, if you have any other question feel free to ask!
Thanks
Stelios
Hello
Examples and bagfiles are posted in https://github.com/mrsp/serow
Although this is a legged robot estimator, the IMU EKF used in it is the same.
Hope it helps Stelios
Hello,
great contribution, thanks. Would you mind posting
to illustrate better how to use your library?
Thanks again, Wojciech