RossHartley / invariant-ekf

C++ library to implement invariant extended Kalman filtering for aided inertial navigation.
BSD 3-Clause "New" or "Revised" License
461 stars 101 forks source link

question about the sensor measurement of kinematic example #9

Open yuan0623 opened 4 years ago

yuan0623 commented 4 years ago

Hi, I'm trying to reproduce your results on my own to make sure that I really understand your paper Contact-Aided Invariant Extended Kalman Filtering for Robot State Estimation*. When I look into the KINEMATIC measurement in imu_kinematic_measurements.txt, I find it is quite strange. Here is my understanding: instead of measuring the joint encoder of the Cassie, the imu_kinematic_measurements.txt provides the relative position between the foot and the robot directly, which I can understand. However, The imu_kinematic_measurements.txt also provides the covariance matrix, which I really don't understand. Can the covariance matrix be measured? This is really strange to me. I would be really appreciated if you can answer this question for me. Thank you so much in advance for your time! Best Yuan

abdul-mannan-khan commented 2 years ago

I am also confused about it. Did you (@yuan0623 ) solve this problem? Please update. Thank you.

RossHartley commented 2 years ago

Apologies for the very delayed response. The covariance matrix was generated by assuming some values of encoder noise and mapping it though the kinematic Jacobian to get the 6x6 matrix.

I did this so the core library does not need the robot-specific kinematic Jacobian.

This was just an example, and I did minimal effort identifying the true encoder noise for Cassie.