Does not work with current OpenSim

[RTnhN]

If you want to try to replicate this without the custom OpenSenseRT RPI image, it will not work. The ikSolver.addOrientationValuesToTrack method was removed in favor of the BufferedOrienationReference.putValues method. Until this PR, there were some problems for the API, but those should be fixed now. You should be able to run it with the following changes to the ik_streaming.py file:

# Was line 97
oRefs = osim.OrientationsReference(orientationsData)

# Is 
oRefs = osim.BufferedOrientationsReference()
rowVecView = orientationsData.getNearestRow(0)
rowVec = osim.RowVectorRotation(rowVecView)     
oRefs.putValues(0, rowVec)

# Was line 97 (actually line 154)
ikSolver.addOrientationValuesToTrack(time_s+dt, rowVec)

# Is
oRefs.putValues(time_s+dt, rowVec)

I can make this into a PR.

[Hari-techi-96]

hi @RTnhN After making these changes to the ik_streaming.py file and As per #3644 i made changes to BufferedOrientationReference.h file, i still got an error

Traceback (most recent call last):
  File "/home/raspi/VR/RealTimeKin/ik_streaming.py", line 101, in <module>
    oRefs.putValues(0, rowVec)
  File "/usr/local/lib/python3.10/dist-packages/opensim-4.5-py3.10.egg/opensim/simulation.py", line 34527, in putValues
    return _simulation.BufferedOrientationsReference_putValues(self, time, dataRow)
TypeError: in method 'BufferedOrientationsReference_putValues', argument 3 of type 'SimTK::RowVector_< SimTK::Rotation > const &'

and ikSolver.addOrintationValuesToTrack(time_s=dt, rowVec) was in line 154.

[RTnhN]

Hi @Hari-techi-96, how are you installing/building opensim? When built, the test suite checks if you can append data to the oRef object: opensim-org/opensim-core@e9c4930

[Hari-techi-96]

Hi @RTnhN i build Opensim in ubuntu 22.04.3 LTS version for my raspberry pi 4B with this instruction. I got suggestions from this repo for troubleshooting. This "test_BufferedOrientationReferencePut()" method was not their in that test_opensense.py file but i added it after going through this link but i didn't run that file after editing. What do i need edit to solve that error?

[RTnhN]

Hi @Hari-techi-96, if you run the shell script to set up OpenSim, there should be no files to edit. It pulls the main branch which has the needed changes. When you "made changes to BufferedOrientationReference.h", you just changed that file, right? You did not do any rebuilding of OpenSim, right? I think the problem is not really in that file, but it is in the interface that SWIG makes based off of that file.

I think the best bet would be to delete any directories that the OpenSim installer made and rerun the script to completely rebuild OpenSim. I know that it is a bummer to wait the 4+ hours for it to build, but I think a clean build is the best way to get it working.

[Hari-techi-96]

@RTnhN Thanks for suggestion I'll try this, can you provide any link where i can get the modified instructions for the New build.

[RTnhN]

@Hari-techi-96 You can use the same instructions that you found here. Just delete the folders that OpenSim created and follow the instructions again.

[Hari-techi-96]

@RTnhN Thankyou so much the changes after the re building of opensim actually worked.

[RTnhN]

Glad to hear @Hari-techi-96 ! Thanks for coming back to this issue and letting us know.

[Hari-techi-96]

hi @RTnhN, the ik_streaming.py ran and it generated raw_imu.npy , timestam.npy and recording.sto files but as per the instructions given in this page the ik_streaming.py file didn't generated the .mot files which is used to visualize the output in opensim.exe . So how should i get the visualization?

[RTnhN]

Hi @Hari-techi-96, I have not gotten that far yet. You do have visualization enabled on line 34, right?

Actually, I think that maybe the instructions are outdated. Mot files are legacy. Maybe you might be able to just use the .sto file.

[Hari-techi-96]

hi @RTnhN sorry for the late replay , i didn't had access to my hardware from the past few days, but today i got the hardware, as you said .mot is legacy and .sto file will work in the opensim . I'm working with 3 IMUs one attached to pelvis and one to left shank and another to left thigh. But in the .sto file i got only one data with timestamp as shown in the below image.

and if i enable visualization on line 34 , the visualizer just pops up and no movement is detected in the simulation.
Button,Pelvis,L_thigh,L_shank, 1,0,9,9, 60 online test_6 500 ISM330DHCX
This is my setting.txt file. i connected the sensor to mux port of 0,6,7 even if i change 1,0,9,9 to 1,0,6,7 same output.

[RTnhN]

Hi @Hari-techi-96, it might be a good idea to test the hardware first to make sure that you are getting the IMU data correctly. I am actually using different hardware, so I can't help much with the specifics.

[Hari-techi-96]

I tested the same hardware with different code for collecting IMUs data , it is working . But when i run this ik_streaming.py i don't get any data in the output. Anyway thanks for the Help.

[RTnhN]

I tested the same hardware with different code for collecting IMUs data , it is working . But when i run this ik_streaming.py i don't get any data in the output. Anyway thanks for the Help.

I actually realized that I am having the same problem too. I think the problem is that the names of the segments is not being loaded into the IK. This will take some time to debug though.

[Hari-techi-96]

can you suggest any alternatives for simulating the IMU data with model?