alro65 5f9b445572 sprint-7: Commissioning wizard + relay ZN auto-tuner + NVS config + knob encoder
Python:
- autotuner.py: RelayAutoTuner (Astrom-Hagglund relay method) → TunerResult
  with Ku/Tu; TunerResult.to_pid_gains() applies ZN formulas
- commissioning_wizard.py: 4-phase state machine (RUDDER_LIMITS → SENSOR_CAL
  → AUTO_TUNE → DONE); transport-injected for testability; abort on invalid cal
- test_autotuner.py: 17 tests covering relay convergence, ZN formulas,
  wizard full-run, abort, ADC swap, identical-ADC guard

Firmware:
- nvs_config.h/cpp: NVS-backed CalibrationData store (adc limits, rudder angles,
  outer Kp/Ki/Kd, commissioned flag); float stored as uint32 via memcpy
- knob_encoder.h/cpp: quadrature rotary encoder on GPIO 16/17 with ISR Gray-code
  decode; knob_arm coil arms for 5 s window; updates heading setpoint ±1 deg/detent
- modbus_slave.cpp: COIL_CMD_KNOB_ARM now calls knob_encoder_set_armed()
- main.cpp: nvs_config_init/load at boot; apply commissioned calibration to
  rudder sensor and outer loop gains; start knob encoder task

Tests: 326 passed | Flash: 28.5%

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-20 02:55:26 -04:00

AR-Autopilot

Professional marine autopilot for vessels in the 30-40 m range (motor yachts, motor sailboats, fishing vessels, small ferries, coastal patrol boats).

Part of the AR Suite alongside AR-ECDIS, VMS-Sailor, AR-ShipDesign, AR-ElecArrangement, and AR-StabCol. Sold standalone or bundled with AR-ECDIS.

NOT Dynamic Positioning. NOT joystick docking. This is a classic heading-and-track autopilot with intelligent drift compensation, controlling rudder actuators (hydraulic or electric).


Status

Sprint 0 — Foundations (in progress).

This sprint delivers the repository structure, core data model, seed library, and a passing test suite. No functional firmware, Studio GUI, or display yet — those start in Sprint 1.

See docs/AR_Autopilot_brief.md for the complete project brief, scope, and roadmap.


Components

Component Tech Purpose
Studio (arautopilot/studio/) Python 3.11 + PySide6 Project configurator (integrator-side, not shipped to customers). Generates per-vessel .appack packages
Firmware (firmware/ar_autopilot_v1/) C++ on ESP32 via PlatformIO Real-time PID control, NMEA 2000 + Modbus, safety logic. Runs on the AR-NMEA-IO v1.0 board (shared with VMS-Sailor)
Display (display/) Flutter Desktop (Win + Linux) Dedicated bridge cockpit-feel touch display with rotary knob input
Core models (arautopilot/core/) Pydantic v2 Shared data model (vessel config, PID config, actuator config, alarms, modes, knob state)
Library (arautopilot/library/) YAML + JSON Curated seed: actuator profiles, default tunings per vessel type

Requirements

  • Python 3.11 or newer
  • Git
  • (Later sprints) PlatformIO, Flutter SDK, WiX Toolset

Quick start (Sprint 0)

# Create venv and install
python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -U pip
pip install -e ".[dev]"

# Run tests
pytest

# Run the Sprint 0 demo (creates, saves, reloads a project config)
python examples/sprint0_demo.py

Repository layout

AR-Autopilot/
├── arautopilot/          # Python package (core models, library, studio stubs, tests)
├── firmware/             # ESP32 firmware (Sprint 1+; only pinout.h in Sprint 0)
├── display/              # Flutter dedicated display (Sprint 4+)
├── examples/             # Runnable demos
├── docs/                 # Brief + per-sprint design docs
├── installer/            # WiX MSI scripts (later)
└── tools/                # Helper scripts (later)

See docs/architecture.md for a one-page architecture overview.


Sprint roadmap

Sprint Focus
0 Foundations: repo structure, core data model, seed library, tests
1 Firmware base (I/O, Modbus, NMEA 2000 read, STANDBY mode)
2 PID inner loop (rudder position control)
3 PID outer loop + Heading Hold (with ROT feed-forward & gain scheduling)
4 Studio + basic dedicated display
5 True Course + Track Keeping (smooth XTE correction)
6 Safety, alarms, NMEA 2000 publish, VMS alarm consumption
7 Knob + commissioning + offline auto-tuning
8 EKF + adaptive tuning + telemetry + VPN
9 Hardening + integrated testing
10+ Phase 2 (wind modes for sailboats) and beyond

Full detail in the brief.


License

Proprietary. All rights reserved. See LICENSE.txt.

Commercial deployment requires a per-vessel license bound to the installation HWID. Contact alro65@gmail.com for licensing.

S
Description
No description provided
Readme 2.2 MiB
Languages
Python 64.7%
C++ 21.5%
Dart 11.7%
C 1.1%
PowerShell 0.6%
Other 0.4%