RRA ridiculous large negative residuals

[AUDE LEFRANC]

Hi All,

I'm hoping you can help me troubleshoot what's going on with my RRA. I've gotten help from everyone in my lab, including the postdoc. No one can identify what is causing the errors. I've also used the troubleshooting page on the website and that didn't help.

When I run RRA, I get ridiculously large negative residuals when running RRA (-80,000 N) which is shown in the attached figure.

Large negative residuals

Residuals RRA.JPG (77.3 KiB) Viewed 597 times

I tried this in both 4.0 and 4.1 and got the same problem. I received very small/good errors when running IK. My actuators and tasks files look good. You can see in the next attached figure that both the marker and force data seems to be ok.

Marker and force data is present

Motion and marker data.JPG (22 KiB) Viewed 597 times

I suspect something is off with the external loads settings. I am applying the left and right force plate forces on the left and right calcaneus, respectively. When I applied the forces to the ground rather than the calcaneus, I got residuals that were all reasonable, besides Fy which was ~+600 N (about the mass of the participant). But I believe the forces should be applied to either foot.

External forces setup 1

ExternalLoads1.JPG (34.63 KiB) Viewed 597 times

External forces setup 2

ExternalLoads2.JPG (42.77 KiB) Viewed 597 times

I should note the following:
- TRC and MOT files were converted from C3D files using the C3Dexport Matlab function
- The error occurs regardless of whether the MOT files are kept as-is, or filtered and NANs removed.
- Data was collected using a split belt treadmill

Thanks so much!

Best,
Aude

[Dimitar Stanev]

Can you provide a plot with the force, point and torque of the ground reaction forces for both legs?

[AUDE LEFRANC]

Sure!

The forces are here:

GroundForces.JPG (55.94 KiB) Viewed 488 times

The CP location is here:

CP no peak.JPG (47.62 KiB) Viewed 488 times

Ground moment:

groundMoment.JPG (78.49 KiB) Viewed 488 times

Best,
Aude

[Dimitar Stanev]

The force components seem OK. The point and torque information is clearly wrong. It seems that there is a lot of noise that compromises the actual information. This is a common problem that should be carefully handled by filtering.

For the point I would expect small values (e.g., 0.2) because it is in meters. The problem manifests when the foot is not in touch with the ground plates (you may observe a lot of jitter). The side effect is that the force is applied at a point that moves rapidly very far away (this can create large moments). If one filters the signal as it is, it will probably distort the information when the foot is in contact due to low signal to noise ration in the region when the foot is not in touch. I would suggest that you remove the information when the foot is not touching the ground and then interpolate the signal. Also, it might be necessary to filter the information afterwards to remove high frequency noise.

As for the torques, it might be a good idea to do the same. Note that 100Nm is a very large torque. One would expect small values 1-10Nm.

Also, make sure to check if the point values are reported in mm. Then you have to convert them to meters. If the point is in mm then the torque is in Nmm, therefore, it should be converted to Nm. For the force it seems that it is in N.

[AUDE LEFRANC]

Hi Dimitar,

Thanks for the insight. So sorry for the delay, I was traveling and didn't have internet access.

I have fixed the point values and moments (attached). When the foot isn't touching the plate, I assigned all CP and moment values to zero. I still have the same problems with the large residuals.

Any other suggestions? I have run my IK results, with the MOT and TRC file overlayed. Everything seems to look fine.

Best,
Aude