Bug between thrusterStateEffector and thrusterDynamicEffector

[MitchellMiya]

Hi all,

I noticed a discrepancy with the implemented force between the thrusterStateEffector module and the thrusterDynamicEffector module. It appears that when accessing the log for thrustForce in the thrOutputMsgPayload using the dynamic effector thruster, the results are incorrect and do not align with those in the thrusterStateEffector. Interestingly, the tracking error still converges to zero for both cases. Please see screenshots attached below.

Note that the implemented force log doesn't exist in the thrusterDynamicEffector example - I just replicated the creation of the log from the thrusterStateEffector example.

Let me know if I should be accessing a different payload message when using the dynamic effector, but please note that the fuelTank module also uses the thrOutputMsgPayload when calculating fuel usage, and thus also produces incorrect results. If there is another message you can recommend to use in the meantime to monitor fuel usage, I'd appreciate it.

Thank you!

Thruster Dynamic Effector results (scenarioAttitudeFeedback2T_TH.py)

Thruster State Effector results (scenarioAttitudeFeedback2T_stateEffTH.py)

[joaogvcarneiro]

Hey Mitchell,

Are you sure you didn't change anything in your first scenario? I've just run both scenarios and they yield results similar to the second example you showed.

[MitchellMiya]

Hi Joao,

Double checked and the only things I have changed are adding the loggers for the force implemented.

Is there any chance I can see how you implemented the logger for the force implemented in the dynamic effector example? I have essentially copied over the logging code from the state effector example (except changing the fig number to avoid conflict) and I'm still getting the incorrect force implemented.

[joaogvcarneiro]

I'm simply running the scenarios within the examples folder.

[MitchellMiya]

Yes - running the scenarios alone will produce similar plots. However, I'd like to emphasize that the produced thrust message is still different between the two - it's just not plotted in the dynamic thruster scenario. I have created this plot as the Force implemented plot in the first image, third plot down to highlight this issue.

While it doesn't make a difference within this scenario, it has affects on any other module accessing the thrustForce message within thrOutputMsgPayload. For example, in our use case, we are seeing incorrect fuel usage because the fuelTank module depends on this message.

Please let me know if there's anything unclear about the way I'm framing this issue. Thanks!

[schaubh]

The two thruster modules to behave slightly differently. The dynamic trusters uses a series of if statement to allow thrust to be turned on instantly, control linear on- and off-ramping, etc. The down side is that the dynamic thruster does not work with variable time step integration as these if statements don't allow the code to roll back time and change the integration time step.

In contrast, the state effector use a low pass filter to process the on/off commands. As a result it is not possible to have an instantaneous jump from 0 to 1 in the thrust state, but the low pass filter frequency will control the smoothed on ramping. The benefit is that this thruster can be used with variable time step integration.

Finally, we just last week discovered a bug in the newer thruster state effectors. Unique default state names were not being set. This means if you added multiple thrusters the last thruster value would over-ride all the other thrust values. This has not been fixed in the latest develop as of a week ago.