ASTRA-Loop

ASTRA-Loop is a Software-in-the-Loop (SITL) simulation environment for testing apogee prediction algorithms and airbrake control strategies on experimental student rocket flight data.

Requirements

• Python 3.10+
• Dependencies listed in requirements.txt

Quick Start

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python main.py

Usage

1. Launch the application with python main.py.
2. Click Browse CSV… and select a flight log (OpenRocket export or Kalman filter log).
3. Choose a Filter, Predictor, and Controller from the dropdown menus.
4. For ApogeePredict3D, optionally choose a Coast Solver (euler or rk4).
5. Click Run Simulation to execute the SITL loop and view plots with performance metrics.
6. For PD calibration, use Calibrate… with data/calibration/scenarios.yaml (see data/calibration/README.md).

See GUIDELINES.md for architecture details and instructions on adding new algorithms.

Apogee predictors

Both predictors estimate apogee during coast after burnout. Physical constants (mass, drag coefficient, reference atmosphere, …) come from Model Parameters.

ApogeePredict1D

Pressure-only estimator for pitot/static-style logs.

Inputs (CSV): static_pressure, total_pressure [Pa]

Pipeline each step:

1. Differential pressure: $$\Delta p = \max(p_{\mathrm{total}} - p_{\mathrm{static}},, 0)$$
2. Altitude from the barometric relation (ISA-style scale height / exponent in ModelParams).
3. Airspeed from Bernoulli using density at the estimated altitude: $$v_z = \sqrt{2,\Delta p / \rho(z)}$$
4. Closed-form coast apogee with quadratic drag. Drag force is taken directly from $\Delta p$ as dynamic pressure: $$F_d = \Delta p \cdot C_d \cdot A$$ so the algorithm does not recompute $\frac{1}{2}\rho v^2$ from the derived speed (avoids redundant calculation and keeps drag consistent with the pressure measurement).

Outputs: position_z, velocity_z, predicted_apogee

When $\Delta p = 0$, apogee falls back to the vacuum ballistic term $z + \frac{m v_z^2}{2mg}$.

OpenRocket pressure/Mach exports can be converted with profile apogee_1d (or apogee_1d_eval if altitude is included for metrics). Total pressure is reconstructed from static pressure and Mach via the isentropic relation: $$P_{\mathrm{total}} = P_{\mathrm{static}},(1 + 0.2,M^2)^{3.5}$$

ApogeePredict3D

State-based coast propagator for Kalman / OpenRocket kinematics logs.

Inputs (CSV): position_z, velocity_z, and either velocity_lateral ($= \sqrt{v_x^2 + v_y^2}$) or both velocity_x and velocity_y

Pipeline each step:

1. Read current $z$, $v_z$, and lateral speed squared $v_{\mathrm{lat}}^2$.
2. Numerically integrate coast dynamics until $v_z \le 0$:

$$ \begin{aligned} \dot{z} &= v_z \\ \dot{v}_z &= -g + k,v_z \\ \frac{d}{dt}(v_{\mathrm{lat}}^2) &= 2,k,v_{\mathrm{lat}}^2 \end{aligned} $$

with

$$ k = -\frac{1}{2m},\rho(z),C_d,A,\sqrt{v_{\mathrm{lat}}^2 + v_z^2}. $$

Outputs: predicted_apogee (final $z$), time_to_apogee

Optimizations / design choices:

• State uses $v_{\mathrm{lat}}^2$ instead of separate $v_x$, $v_y$ — one less variable in the ODE, same physics for axisymmetric drag.
• Shared derivative function _coast_derivatives for both integrators.
• Chooseable solver in the GUI (Coast Solver, only when this predictor is selected):
  • euler — forward Euler (default, cheaper per step)
  • rk4 — classical 4th-order Runge–Kutta (more accurate for the same dt)

Default step size is $dt = 0.01,\mathrm{s}$.