ArduPilot AI-Assisted Log Diagnosis_

Project Overview

This repository serves as a technical proof-of-concept for the AI-Assisted Log Diagnosis project under ArduPilot for GSoC 2026. The goal is to bridge the gap between complex binary telemetry data and human-readable flight insights.

The Demo

The current script (demo_analyzer.py) demonstrates a core logic engine that:

1. Scans Telemetry Streams: Monitors altitude, vibration levels, and system messages.
2. Identifies Anomalies: Uses threshold-based logic to detect mechanical stress (Vibration) and sensor failures (EKF).
3. Generates Diagnosis: Provides clear, actionable feedback for the pilot.

Tech Stack

• Language: Python 3.x
• Environment: GitHub Codespaces
• Libraries: pymavlink (in progress), json, time

Development Roadmap

• Initial Prototype (Mock Logic)
• Integration with pymavlink for .tlog parsing
• Classification model for "Crash vs. Pilot Error"
• CLI Tool for automated batch log processing

Demo Execution

Below is the output of the demo_analyzer.py script running in GitHub Codespaces, successfully identifying flight anomalies:

Contributor

Shaik Rafi - 2nd Year AIML Student
Mentor: Nate Mailhot