AeroPlanX — Unified Mission Planning & Simulation Framework

Graduate-level aerospace systems engineering | AeroHack submission

A single, unified, constraint-based mission planning engine that plans and simulates:

• ✈ Aircraft missions — UAV/fixed-wing waypoint flight under wind, energy, maneuver, and geofence constraints
• 🛰 Spacecraft missions — 7-day CubeSat LEO observation and downlink scheduling under visibility, power, and pointing constraints

One Planner.solve() loop. Zero constraint violations. Fully reproducible.

# 1. Install dependencies
pip install numpy scipy matplotlib

# 2. Run everything with one command (from project root)
cd c:\xampp\htdocs\AeroPlanX
python run.py

# 3. Open the dashboard
# Start XAMPP, then visit: http://localhost/AeroPlanX/frontend/index.html

AeroPlanX/
├── run.py                          ← Single command entry point
├── planner/
│   ├── core/
│   │   ├── state.py                ← Unified MissionState abstraction
│   │   ├── action.py               ← Action types (maneuver, observe, downlink, idle)
│   │   ├── constraints.py          ← All hard constraints (explicit, audited)
│   │   ├── objective.py            ← Scoring functions
│   │   └── planner.py              ← UnifiedPlanner.solve() — shared solver loop
│   ├── aircraft/
│   │   ├── wind.py                 ← Dryden turbulence + spatial wind model
│   │   ├── model.py                ← Fixed-wing UAV dynamics & energy model
│   │   └── mission.py              ← Aircraft mission orchestrator
│   ├── spacecraft/
│   │   ├── orbit.py                ← Two-body + J2 LEO propagator
│   │   ├── visibility.py           ← Ground target & station visibility windows
│   │   └── mission.py              ← 7-day spacecraft scheduler
│   └── validation/
│       ├── monte_carlo.py          ← Monte-Carlo wind robustness (N=50 runs)
│       └── metrics.py              ← Metrics export + constraint audit
├── backend/
│   ├── api.php                     ← REST API for mission data
│   └── db.php                      ← MySQL schema
├── frontend/
│   ├── index.html                  ← Dashboard (Overview/Aircraft/Spacecraft/Validation)
│   ├── style.css                   ← Aerospace dark theme
│   └── app.js                      ← Dashboard data loading
├── outputs/                        ← Generated plots + metrics (gitignored)
└── docs/
    └── report.md                   ← Technical report

1. Aircraft Mission — Plans a 10-waypoint UAV survey mission with real wind physics, saves trajectory + energy plots to /outputs/
2. Spacecraft Mission — Plans a 7-day CubeSat scheduling mission, saves timeline, energy, ground track plots + schedule CSV
3. Monte-Carlo Validation — Runs 50 independent aircraft missions with different wind seeds, compares planner vs greedy baseline, saves statistical plots
4. Summary Export — Writes /outputs/summary.json (consumed by the dashboard)

1. True Unified Architecture: Not two scripts glued together, but a single mathematical core (UnifiedPlanner) solving both aircraft and spacecraft missions.
2. Explicit, Auditable Constraints: Every safety rule is a distinct, testable Python class. Zero "black box" logic.
3. Robust Under Uncertainty: Validated against 50 randomized wind fields with 100% success rate, outperforming standard baselines.
4. Reproducible Engineering: Deterministic seeding, config-as-code, and one-command execution.
5. Aerospace Systems Thinking: Derived from real physics (Dryden turbulence, J2 perturbations) rather than game-like approximations.

plan = UnifiedPlanner().solve(
    initial_state=state,
    candidate_generator=mission.generate_candidates,  # domain-specific
    constraints=ConstraintChecker([...]),             # explicit, audited
    objective=MinTimeEnergyObjective(),               # domain-specific scoring
    terminal_condition=lambda s: s.payload["wp_idx"] >= len(waypoints),
)

The same UnifiedPlanner.solve() runs both aircraft and spacecraft missions. Domain specifics are injected — the core algorithm never changes.

All constraints are in planner/core/constraints.py. Every check returns (bool, str) — no silent passes, no implicit logic. The planner maintains a full audit trail.

The current implementation is fully deterministic. AI integration can be added by injecting AI-suggested candidate actions through the candidate_generator — the constraint system will filter them deterministically.

After running python run.py, the /outputs/ directory contains:

With XAMPP running, visit: http://localhost/AeroPlanX/frontend/index.html

The dashboard automatically loads from /outputs/summary.json via the PHP API and auto-refreshes every 30 seconds.

# Check constraint violations (must be 0)
python -c "import json; d=json.load(open('outputs/summary.json')); print('Violations:', d['total_constraint_violations'])"

# Check Monte-Carlo success rate (must be 1.0)
python -c "import json; d=json.load(open('outputs/monte_carlo_results.json')); print('MC Rate:', d['success_rate'])"

• Wind: Dryden turbulence (MIL-HDBK-1797) + sinusoidal spatial field
• Dynamics: V_ground = V_airspeed + V_wind, coordinated turn ω = g·tan(φ)/V
• Energy: E_rate = (P_avionics + P_cruise·pf²) / E_capacity

• Orbit: Two-body + J2 secular perturbation, Kepler equation (Newton-Raphson)
• Visibility: Elevation angle geometry in ECI frame → time windows
• Power: Normalised duty-cycle budget (≤40% per orbit)