Optisat AI

🛰️ OptiSat-AI: Autonomous Optical Satellite Mission Planner

🌌 Inspiration

In traditional Earth Observation (EO) satellites, the mission planning process requires engineers at ground stations to perform complex mathematical calculations and manual coordinate entries. During a natural disaster (floods, wildfires, etc.), this process can lead to critical time losses. OptiSat-AI was designed to eliminate this operational burden and fully automate Earth observation missions by combining orbital mechanics with modern AI capabilities.

⚙️ What it does

OptiSat-AI converts complex, natural language mission commands from users or ground operators directly into numerical and visual orbital telemetry. The system processes text-based commands to determine the target region's geographic coordinates, calculates the optimum optical camera off-nadir (roll) angle required for the satellite to lock onto that region, and dynamically simulates this on an interactive 3D Earth model.

🧠 How we used the Gemini API

In this project, the Gemini API goes beyond being a standard chatbot; it is positioned as the core "Autonomous Mission Engineer" (AI Agent) of the system:

1. Text-to-Telemetry: Location and analysis requests entered as free text by the user are processed by Gemini.
2. Structured JSON Output: The Gemini API was forced to produce output in a strict engineering format using the response_mime_type: "application/json" parameter. The latitude, longitude, and roll_angle data returned by the API are read directly by the 3D Plotly physics engine on the frontend to autonomously drive the map.

🛠️ How we built it

• Backend & Frontend: Developed a fast and stable interface using a Python-based Streamlit architecture.
• 3D Visualization: Integrated a global (orthographic) Earth model using the Plotly Graph Objects library.
• Safe Presentation Mode (Fallback Engine): To counter API quota limits (HTTP 429)—the biggest risk during hackathons—we wrote a smart, offline fallback engine. If the API limit is reached, the system continues to run seamlessly without crashing.
• Deployment: The project was made entirely open-source on GitHub and deployed live on the Streamlit Community Cloud.

🏆 Challenges we ran into

Getting the API to consistently produce perfectly accurate coordinates and angles in a short time was initially challenging, but we overcame this with precise prompt engineering and forced JSON schemas. Additionally, by developing a "Simulation Mode" against potential limitations during live presentations, we experienced firsthand how critical fault tolerance (fail-safe mechanisms) is in software architecture.

🚀 What's next

• Integrating SGP4 orbital propagation algorithms to calculate the satellite's real-time pass times.
• Making link budgets and power consumption optimizations calculable by the AI.

Built With

• python