Skip to content

README.md

Latest commit

 

History

History
146 lines (107 loc) · 6.91 KB

README.md

File metadata and controls

146 lines (107 loc) · 6.91 KB
Raw

RAWES Simulation

End-to-end flight simulation for the Rotary Airborne Wind Energy System (RAWES) — a 4-blade tethered autorotating rotor kite controlled via trailing-edge flaps through an H3-120 swashplate.

Combines ArduPilot SITL (flight controller logic), a Python RK4 dynamics engine, and a Python mediator (aerodynamics + sensor simulation).

Architecture

┌─────────────────────────────────────────────────────────────────┐
│ ArduPilot SITL (heli JSON backend)                              │
│   Sends servo PWM commands, receives JSON sensor state          │
└────────────────┬──────────────────────────────────────────────┘
                 │ UDP 9002 (servo commands)
                 │ UDP 9003 (sensor state JSON)
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│ mediator.py  (400 Hz loop)                                      │
│   servo PWM → swashplate.py → collective, tilt_lon, tilt_lat   │
│   aero.py        → aerodynamic wrench (NED world frame)        │
│   tether.py      → tether tension force + moment               │
│   dynamics.py    → RK4 6-DOF step → {pos, vel, R, omega}       │
│   sensor.py      → build_sitl_packet() → NED sensor outputs    │
└─────────────────────────────────────────────────────────────────┘
                 │ MAVLink TCP 5760
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│ gcs.py / test GCS                                               │
│   Arm, set mode, send RC overrides, read MAVLink messages       │
└─────────────────────────────────────────────────────────────────┘

Module Summary

File Role
mediator.py 400 Hz co-simulation loop — orchestrates all subsystems
dynamics.py Python RK4 6-DOF rigid-body integrator (gravity internal)
aero.py De Schutter (2018) BEM aerodynamic model
tether.py Tension-only elastic tether (Dyneema SK75)
swashplate.py H3-120 inverse mixing and cyclic blade pitch math
frames.py T_ENU_NED legacy constant and build_orb_frame() — shared coordinate-frame utilities
sensor.py build_sitl_packet() — NED truth state → ArduPilot JSON sensor packet
sitl_interface.py ArduPilot SITL UDP binary protocol (servo recv, state send)
controller.py compute_swashplate_from_state() — truth-state tether-alignment controller; compute_rc_from_attitude() — ACRO RC override controller
gcs.py MAVLink GCS client (arm, mode, RC override, params)
flight_report.py Multi-panel PNG flight report from position/attitude/servo history

Analysis scripts (not part of simulation runtime): analysis/

Coordinate Frames

All defined in and imported from frames.py.

Frame Axes Where used
NED (world) X=North, Y=East, Z=Down dynamics, aero, tether, controller, sensor, SITL
Body columns of R_hub (body→world) gyro, accel, swashplate

All physics uses NED. T_ENU_NED in frames.py is kept as a utility for converting legacy ENU data.

Sensor Design — Physical Attitude

PhysicalSensor in sensor.py reports the true physical attitude orientation (~124° roll / −46° pitch at tether equilibrium, ZYX Euler NED). ACRO mode is used because it only damps angular rates, so the large physical tilt causes no automatic corrective cyclic.

  • rpy = actual ZYX Euler angles from R_hub directly (no overrides)
  • Gyro: R_hub.T @ omega_body — full body angular velocity; GB4008 keeps body non-rotating via K_YAW dynamics, no sensor stripping
  • Accel: specific force R_hub.T @ (accel_world − gravity) in body frame

Initial State

Default starting state (warmup-settled equilibrium, 50 m tether at ~30° elevation):

Parameter Value
pos (NED) [14.241, 46.258, -12.530] m
vel (NED) [0.916, -0.257, 0.093] m/s
body_z [0.851, 0.305, 0.427] (axle aligned with tether)
omega_spin 20.148 rad/s

Regenerate: pytest tests/unit/test_steady_flight.py → writes steady_state_starting.json.

Running Tests

Unit tests (Windows, no Docker)

bash test.sh unit -q

Key unit tests:

  • test_closed_loop.py -- closed-loop physics (dynamics + aero + tether + controller), no ArduPilot
  • test_steady_flight.py -- open-loop equilibrium, writes steady_state_starting.json
  • test_controller.py -- controller function unit tests

Stack integration tests (Docker via WSL)

bash test.sh stack -v

See CLAUDE.md in the repo root for full Docker setup and test commands.

Analysis Tools

Standalone scripts in analysis/. Not part of the simulation runtime; not imported by mediator.py or any test fixture. Run after stack tests to produce structured reports and plots.

Always run analyse_run.py after a stack test:

# List available test runs (newest first)
bash test.sh exec 'python3 /rawes/simulation/analysis/analyse_run.py'
# Analyse a specific test
bash test.sh exec 'python3 /rawes/simulation/analysis/analyse_run.py test_acro_armed'
bash test.sh exec 'python3 /rawes/simulation/analysis/analyse_run.py test_pumping_cycle --plot'
Script Purpose Status
analyse_run.py Post-run structured report; reads logs/{test_name}/telemetry.csv + mediator.log Active
analyse_gps_fusion.py EKF3 GPS fusion event analysis from logs/{test_name}/gcs.log Active
analyse_pumping_cycle.py Pumping cycle energy/tension analysis from telemetry CSV Active
generate_flight_report.py Offline multi-panel PNG plot from mediator telemetry CSV Active
merge_logs.py Merge mediator/SITL/GCS logs in timestamp order for unified timeline Active
build.py Docker image build progress monitor Active
compare_rotors.py Side-by-side rotor definition comparison Active
sg6042_polar.py SG6042 airfoil polar plotter Active

Per-test logs: simulation/logs/{test_name}/ -- telemetry.csv, mediator.log, sitl.log, gcs.log, arducopter.log. Suite log: simulation/logs/pytest_last_run.log.