JSBSim yielding odd results

[Jolyon-S]

Hey folks,

I'm working on a setup with ArduPilot SITL and JSBSim, and I've found that the outputs from each step of JSBSim seem at times pretty weird.

For reference, the model I'm running with is the Rascal110 - I'm not dead set on this model in particular though, as I just want to simulate an RC plane (i.e. one that has four PWM controls - aileron, elevator, throttle, rudder, assuming I'm reading the outputs from ArduPilot correctly).

My setup involves:

1. Receiving the latest control update from ArduPilot: this is the controls for JSBSim
2. Updating the controls in JSBSim, through something like:

    this->fdmex.GetFCS()->SetDaCmd(update.aileron);
    this->fdmex.GetFCS()->SetDeCmd(update.elevator);
    this->fdmex.GetFCS()->SetThrottleCmd(0, update.throttle);
    this->fdmex.GetFCS()->SetDrCmd(update.rudder);

3. Running one step of the iteration (dt = 0.001, and I've run DoTrim(2) beforehand but no other initial setup such as setting initial conditions)
4. after step, send the updated information to ArduPilot (e.g. retrieving through this->fdmex.GetPropagate()..., using the ArduPilot JSON API).

Then, my observations are:

1. When the control signals are all 0.0, the plane remains pretty steady (great)
2. When ArduPilot decides to try and take off, it starts increasing throttle and the elevator (also seems reasonable). I'm not sure if elevator --> 1.0 means the plane is trying to go up or down (bit of a plane noob) but that's what is going on
3. The position vector typically starts at 0.0, goes to around 6.0 (which I take to mean I'm 6m underground?), then starts ascending. Weirdly, I get GEAR CONTACT messages while I'm apparently in the air (as high as 100m), so not sure what happens there. After some time I get a TAKEOFF message and my height briefly shows 15m or so again, then snaps back to around 120m - some time afterwards, I get a crash message then the simulation ends.
4. [EDIT]: I've noticed that my position from GetPropagate()->GetLocation(1) stays the same throughout my simulation, even while the other values change (as I'd expect). I doubt my aircraft is so precise that it flies on such a perfect straight line; perhaps this is part of why this is failing?

Does anyone have any idea what is causing my peculiar simulation results? As mentioned, if the suspicion is the model I'm open to trying alternatives, so long as they're relatively simple to fly (ideally with roughly the same 4 inputs).