_____
  / ___ \
 / /   \ \   GOOSE
| |     | |  Flight Forensics & Investigation Platform
| |     | |
 \ \___/ /   Local-first. Evidence-preserving. Profile-aware.
  \_____/

---

Goose is an open-source drone flight forensics platform — not just a crash analyzer. It turns ULog, DataFlash, TLog, and CSV flight logs into structured case files with evidence chains, ranked hypotheses, and PDF/JSON exports. The platform pairs a FastAPI backend with a React 19 frontend, ships 16 built-in analysis plugins covering crash detection through environmental inference, and includes a full case management system with chain-of-custody audit trails — all running locally, with nothing leaving your machine. Goose Pro extends it with GPS-denied navigation validation, MIL-STD-810H compliance reporting, fleet management, and RBAC via Core's public seams.

---

Supported Log Formats

Format	Notes
ULog (.ulg)	PX4 firmware — full telemetry extraction across 20+ streams
DataFlash (.bin, .log)	ArduPilot firmware — message-level extraction
TLog (.tlog)	MAVLink telemetry stream — Core stub; full parser in Goose Pro
CSV (.csv)	Generic tabular logs — stream heuristics for column mapping

---

16 Core Plugins

Crash & Impact

Plugin	What It Does
crash_detection	Detects crash events and classifies root cause (motor failure, power loss, GPS loss, impact, mechanical, pilot error) with probabilistic confidence scoring
failsafe_events	Catalogs every failsafe trigger, emergency mode transition, and classifies failsafe vs pilot-initiated actions
damage_impact_classification	Classifies post-impact damage signatures from IMU and attitude data

Power

Plugin	What It Does
battery_sag	Monitors voltage sag under load, per-cell thresholds, brownout risk, and temperature
log_health	Validates log integrity, stream coverage, sensor dropout, and data-rate continuity

Navigation

Plugin	What It Does
gps_health	Tracks fix quality, satellite count, HDOP, position drift, dropout events, and jamming indicators
ekf_consistency	Evaluates EKF innovation ratios, magnetometer consistency, reset events, and estimator fault flags
position_tracking	Measures horizontal position error (Haversine), vertical altitude error, and hover drift

Mechanical

Plugin	What It Does
vibration	Computes IMU vibration RMS/peak per axis, accelerometer clipping counts, and bearing degradation trends
motor_saturation	Detects motor output saturation, cross-axis asymmetry, failure signatures, and headroom
attitude_tracking	Measures roll/pitch/yaw tracking error vs setpoint and detects PID oscillation

Operator & Comms

Plugin	What It Does
rc_signal	Monitors RSSI, dropout duration, stuck channels, and RC failsafe triggers
link_telemetry_health	Assesses telemetry link quality, dropout events, and RSSI degradation trends
operator_action_sequence	Reconstructs operator input patterns — mode switches, arm/disarm cycles, RC command sequences

Environment & Payload

Plugin	What It Does
environment_conditions	Infers wind speed, direction, and GPS multipath indicators from flight telemetry
payload_change_detection	Identifies candidate mid-flight payload or mass-change events from dynamics signatures
mission_phase_anomaly	Flags anomalous behavior within detected mission phases (takeoff, cruise, landing)

---

ML / Forensics Engine

• Anomaly scoring — weighted confidence scoring across multiple signal dimensions; each finding carries an explicit confidence value scoped to its evidence window
• Ranked root cause hypotheses — a hypothesis engine aggregates plugin findings into ranked causal explanations with supporting and contradicting evidence chains
• Canonical finding normalization — all plugins emit ForensicFinding objects with EvidenceReference links; findings are never raw strings
• Validation corpus — ground truth labels and a structured test corpus back the anomaly scoring thresholds
• Per-vehicle tuning profiles — seven data-driven profiles (Racer, Research, Shop/Repair, Factory/QA, Gov/Mil, Advanced, Default) adjust plugin emphasis and report wording without touching forensic truth

---

Installation

pip install goose-flight

goose --help
goose analyze <file>

To verify a fresh install:

goose doctor

---

Web UI

goose serve

Opens at http://localhost:8080. Subsystem pages:

• Motors — per-motor output, saturation, asymmetry
• Battery — voltage, current, sag events
• GPS / Nav — fix quality, HDOP, dropout map
• Vibration — IMU RMS/peak, clipping counts
• Control — attitude / rate / position tracking error
• Environment — wind estimate, multipath indicators
• Flight Path 3D — interactive 3D trajectory from GPS + altitude
• Anomaly Timeline — all findings across the full flight timeline

Additional views: case management, evidence chain with SHA-256 hashes, hypothesis ranking, and PDF/JSON report export.

---

Extending Goose

Goose exposes four public seams for adding capabilities without forking Core:

Seam	How to Use
goose.plugins entry point	Register analysis plugins via pyproject.toml entry_points
register_parser()	Add support for new log formats
register_report_generator()	Add new report output types (PDF, HTML, JSON, etc.)
register_capability()	Declare optional capabilities that the UI and CLI can discover

Goose Pro uses these seams. Core never imports Pro. Any extension that registers through these seams is automatically discovered at startup — no Core edits required.

---

Goose Pro

Goose Pro is a paid local-install extension that adds GPS-denied navigation validation (CEP/R95/R99 accuracy metrics, NAV-001 through NAV-012 standardized test matrix, 20 dedicated plugins), RINEX 2.x/3.x ground truth ingestion with ECEF-to-WGS-84 conversion, MIL-STD-810H compliance reports, fleet management with per-drone nav system profiles, and role-based access control. See flygoose.dev/pro.

---

Architecture Boundary

Goose Pro extends Goose Core via Core's public seams. Core never imports Pro. Installing Pro alongside Core is additive — uninstalling returns Core to its open-source behavior with zero residue.

Log File (.ulg / .bin / .log / .tlog / .csv)
     |
     v
  [ Parser Framework ]   <-- register_parser() seam
     |
     v
 [ ParseResult: Flight + ParseDiagnostics + Provenance ]
     |
     v
 [ Plugin Engine ]  <-- goose.plugins entry_points
  |    |    |    |    ... (16 Core + N Pro plugins)
  v    v    v    v
 [ ForensicFinding + EvidenceReference ]
     |
     v
 [ Hypothesis Engine ]
     |
     +--> [ Quick Analysis session ]   (in-memory)
     |
     +--> [ Investigation Case ]  -->  [ Audit Log ]
              |
              v
       [ Web UI / CLI ]
              |
              v
       [ Report Generator ]   <-- register_report_generator() seam

---

License

Apache 2.0. See LICENSE.