Add theoretical mode #374
Add theoretical mode #374
Conversation
|
The input boxes don't fit. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
1c47002
to
4e61434
Compare
|
Should be fixed. It'd be nice if I could reliably reset the formatting to default - we should have a button for that. |
|
The response timescale box is missing from the JSON analysis. I'd expect the non-JSON box layout to just be a subset of the JSON box layout. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
4e61434
to
487b0bd
Compare
|
Have you tested the data loading part (seems to get blocked on plotting graphs)? For some reason sysid just freezes for me when I try to load data. My computer has been acting weird recently so that might be the issue, not the code itself. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
487b0bd
to
cfba407
Compare
|
It seems like data errors causes none of the graphs to be plotted. |
|
That is correct; I can change that if you want. |
|
Yeah ideally they can see the raw data so they can fix the motion threshold |
Oblarg
force-pushed
the
theoretical-mode
branch
from
cfba407
to
b8a7de6
Compare
There was a problem hiding this comment.
One concern is that the weird value update bug happens if we load the file the second time:
I haven't checked if this happens on main but it's probably some logic we can fix on this PR.
https://user-images.githubusercontent.com/29788153/154989342-ba430192-55ff-410d-96d8-2f640336ca10.mp4
Edit: The video came out blurrier than I expected but essentially the first time around the calculated gain is at 2.0879 and then the second time around it starts at 2.0879 and then pressing on convert gains to encoder counts turns it into 2.7182
This happens in main immediately upon file load; I'm not sure why loading it twice would cause issues, though... |
For some reason, I can't reproduce it on main. Sounds like some sort of UB then. |
|
I'm having a hard time replicating the bug at all, even on this branch - I think something might be funky here. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
b8a7de6
to
0788f24
Compare
|
Found the offending code, and it's fixed. Should be good now. |
|
what was the error? |
|
Failing to early exit out of the render function in a data error state if the reset button is pressed. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
0788f24
to
d444ff7
Compare
|
The error still shows up for me :/ |
Oblarg
force-pushed
the
theoretical-mode
branch
from
d444ff7
to
94978d8
Compare
|
What dataset is causing it? I cannot replicate it. |
|
The original one linked on the discord. I'm assuming it might be related to the convert to encoder gains checkbox. |
It would be good to upload it here. |
Oblarg
force-pushed
the
theoretical-mode
branch
from
94978d8
to
b57babc
Compare
Replaces the "waiting for data json" state with a theoretical analysis mode that allows you to plug in your own calculated feedforward gains for use with LQR. Also refactors the state machine to get rid of states that do nothing but call code and transition to other states (these are now just subroutines).