Manually changing spacecraft states mid-simulation

[atharris-work]

I have reason to manually move spacecraft around (i.e., change their position and velocity) mid-simulation. While there is an example of this being done with state references, this approach results in strange SWIG/Eigen errors when multiple spacecraft, each with multiple effectors, exists:

 pos_ref.setState(new_pos) # new_pos is a (3,) numpy array
  File "/basilisk/dist3/Basilisk/simulation/MtbEffector.py", line 356, in setState
    return _MtbEffector.StateData_setState(self, newState)
TypeError: in method 'StateData_setState', argument 2 of type 'Eigen::MatrixXd const &'

As an alternative, I tried manually changing the r_BN_N_Init and v_BN_N_Init vectors and then calling reset on the spacecraft; however, this produces a plethora of BSK_WARNING messages about creating new states with the same names as old states.

What is the correct or preferred way to change spacecraft state values mid-sim?

[Mark2000]

I recently was attempting to do the same, and this works for me in complex multi-spacecraft sims for instantaneous changes in position and/or velocity: https://github.com/AVSLab/bsk_rl/blob/0a4a9d09f1c58d8ca8d91fb400f6a7c89c8c6328/src/bsk_rl/sim/fsw.py#L998C1-L1008C1

In short:

scObject.dynManager.getStateObject(scObject.hub.nameOfHubPosition).setState([r_x, r_y, r_z])
scObject.dynManager.getStateObject(scObject.hub.nameOfHubVelocity).setState([v_x, v_y, v_z])

It's pretty similar to that example script, but I don't think its exactly the same.

[atharris-work]

I'll try to run your example later, but I suspect something is more...deeply broken if calling setState requires a call through simulation/MtbEffector.py.

[atharris-work]

Following up on this, even just importing MtbEffector in scenarioHohmann.py also causes the stateData error:

# Line 96, imports
from Basilisk.simulation import reactionWheelStateEffector, simpleNav, spacecraft, ephemerisConverter, MtbEffector # Only place MtbEffector is referenced in the script

Running results in:

  File "/home/andrew/workspace/basilisk/examples/scenarioHohmann.py", line 511, in <module>
    run(
  File "/home/andrew/workspace/basilisk/examples/scenarioHohmann.py", line 419, in run
    velRef.setState(new_v)
  File "/home/andrew/workspace/basilisk/dist3/Basilisk/simulation/MtbEffector.py", line 356, in setState
    return _MtbEffector.StateData_setState(self, newState)
TypeError: in method 'StateData_setState', argument 2 of type 'Eigen::MatrixXd const &'

I'm guessing this is a SWIG error at some level with MtbEffector? It seems like it overrides the other StateData/DynManager classes when imported.

Environment specifics:

• OS: Ubuntu 22.04
• Python: 3.10.12
• BSK Version: Most recent develop, replicated on older version as well

[atharris-work]

Seems like this can be fixed by adding

%include "swig_eigen.i"

to MtbEffector.i - doing this, rebuilding, and re-running sceanrioHohmann.py with an unused MtbEffector.py import runs successfully.

[atharris-work]

Seems straightforward enough to fix, created issue #850 to track.