[paparazzi-commits] [paparazzi/paparazzi] ca1414: Optical flow filter and derotation fix (#1855)

[GitHub]

  Branch: refs/heads/master
  Home:   https://github.com/paparazzi/paparazzi
  Commit: ca141420eb06086b75fd4a9cf31e39da8038e1a3
      
https://github.com/paparazzi/paparazzi/commit/ca141420eb06086b75fd4a9cf31e39da8038e1a3
  Author: knmcguire <address@hidden>
  Date:   2016-09-29 (Thu, 29 Sep 2016)

  Changed paths:
    M conf/airframes/TUDELFT/tudelft_ardrone2_opticflow_stereo.xml
    M conf/airframes/TUDELFT/tudelft_bebop_opticflow.xml
    M conf/modules/cv_opticflow.xml
    A sw/airborne/modules/computer_vision/lib/filters/kalman_filter_vision.c
    A sw/airborne/modules/computer_vision/lib/filters/kalman_filter_vision.h
    M sw/airborne/modules/computer_vision/opticflow/inter_thread_data.h
    M sw/airborne/modules/computer_vision/opticflow/opticflow_calculator.c
    M sw/airborne/modules/computer_vision/opticflow/opticflow_calculator.h
    M sw/airborne/modules/computer_vision/opticflow_module.c

  Log Message:
  -----------
  Optical flow filter and derotation fix (#1855)

Here is some improvements to the optical flow code.

-  Kalman filter: I made a generic kalman filter just for the use of vision. 
The problem of using the HHF filter is that it also effect the reaction of the 
controls (for some reason there is too much dependency on the process model but 
I can't change the noise factors). I think that it is wise to filter vision 
results separate from the state filters. Right now it is implemented for 
optical flow, but later it can be used for tracking as well. I hope you will 
agree
 -  derotation correction factor: there seems to be an error between the 
calculated derotation flow by the gyro and the calculated flow. Now it has been 
fitted together and for the bebop a correction term has to be added (credits to 
@mmoiozo for figuring this out).

This has been test flown with two bebops.