search

ethz-asl / rio

Graph-based, sparse radar-inertial odometry estimation
BSD 3-Clause "New" or "Revised" License
81 stars 5 forks source link

Estimations increase dramatically #7

Open cmakelabs opened 1 month ago

cmakelabs commented 1 month ago

While rio runs well on the provided datasets, however when I try to launch it with my own rosbag, the estimate position and velocity seem to increase very rapidly, and converge to wrong numbers, as the robot real speed is 0.5 m/s (while the ones shown below will increase indefinitely as the bag is playing) :

pose: 
  pose: 
    position: 
      x: -160.44983290133843
      y: -14.600831309434668
      z: -47.69175573838993
    orientation: 
      x: -0.02611747507882512
      y: -0.032748962918190724
      z: 0.9235699756457728
      w: 0.38113471154690987
  covariance: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
twist: 
  twist: 
    linear: 
      x: 8.807195182027714
      y: 12.875944052624204
      z: -2.5869651071525483
    angular: 
      x: 0.05528554320335389
      y: -0.012584030628204346
      z: 0.0011943281861022117
  covariance: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
---

There is also this series of warnings, that I could not debug properly, as the timings seem to be too close and the mismatch is only in the last fractions of time, not sure if this what causing the wrong estimation:


[ WARN] [1721221984.697609407]: Failed to split propagation, skipping CFAR detections.
[ WARN] [1721221984.697652414]: Split time: 1695304190.525563847
[ WARN] [1721221984.697669591]: Last IMU time: 1695304190.519120621
[ WARN] [1721221984.797759724]: Failed to split propagation, skipping CFAR detections.
[ WARN] [1721221984.797818564]: Split time: 1695304190.625555836
[ WARN] [1721221984.797849152]: Last IMU time: 1695304190.618970971
[ WARN] [1721221984.897949333]: Failed to split propagation, skipping CFAR detections.
[ WARN] [1721221984.898010507]: Split time: 1695304190.725679948
[ WARN] [1721221984.898040557]: Last IMU time: 1695304190.718575103

Would sincerely appreciate your answer

rikba commented 1 month ago

Hi there, most likely it diverges when only IMU messages are considered. Did you implement the buffer https://github.com/ethz-asl/rio/issues/6 ?

cmakelabs commented 1 month ago

Have not got time unfortunately, will ping you with pull-request once I got time to do it

rikba commented 1 month ago

I think a buffer will solve the problem with early arriving radar messages. I may have time to implement it at some point as well. You can contact me by mail at brik@ethz.ch, and I can look at your data set.

cmakelabs commented 1 month ago

Thank you very much, I have sent you an email with the dataset attached

Dcasadoherraez commented 5 days ago

Any updates on the buffers @rikba?