Sys ID Simulation Data Way off from rest of Data

Hey! Our team is new to robot characterization and we’ve been trying to get Sys ID to work on test bot recently. However, we seem to be having trouble with the way that we’re entering parameters or the way it’s plotting the simulation data.

Either way it’s causing our sim velocity r squared value to be incredibly low, which is making us unable to use the constants we’re being given (to my knowledge).

Also if it’s helpful to know, our encoders do seem to giving the proper distances back when running the tests.

Here’s a google drive folder with images of our graphs and inputted parameters: Sys ID Graphs and Parameter Inputs - Google Drive

Let me know if y’all need more information, I’ll be glad to give it :slight_smile:

Your gearing appears to be off. Are you sure your DT is geared 1:1?

1 Like

Good point, I’ll take a closer look and do some further testing with that in mind. Let you know how it goes.

Upon further testing it doesn’t appear to be an issue with our gear ratio, in fact the recorded encoder distance seemed to off by a factor of whatever we put in as the gear ratio.

It seems to be more of an issue of the “analysis type” being set as “drivetrain” in the generator, at least for characterizing our robot’s movement on the x axis.

What I mean by that is we do seem to be getting better plot simulation data when we set our mechanism on the logger as “drivetrain (angular)” while it’s listed as “drivetrain” in the generator as shown in the image below:

I was able to find a work around to what seems to be good data for our robot’s movement on the x axis while setting the analysis type as “general mechanism” on the generator and “simple” on the logger as shown in the image below:

I’m not sure why this work around works, however, and will be trying to narrow down things and hopefully finding out why.

The “drivetrain” and “drivetrain (angular)” both use the drivetrain generator type. The non-angular logger setting runs the robot in a straight line (both wheels going the same direction), while the angular version spins the robot in place (wheels going in different directions. This is to achieve analysis of both linear and angular velocity and acceleration.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.