AstroTEA: Space Agriculture AI

AI-powered decision engine for space agriculture, optimizing crops, resources, and sustainability under mission constraints.

Comment

AstroTEA: Space Agriculture AI

Inspiration

Long-duration space missions face a fundamental challenge: how to sustain life with limited resources.

Traditional planning methods are often static and fail to adapt to dynamic conditions such as resource depletion, system failures, or environmental variability. The goal of this project was to explore whether we could build a system that behaves like a mission planner while adapting dynamically like a living system.

This led to the creation of AstroTEA, a decision intelligence system for space agriculture.

What It Does

AstroTEA is designed to:

  • Select optimal combinations of crops, algae, and microbial systems
  • Allocate limited resources such as water, energy, and space
  • Simulate how these systems behave over time
  • Provide clear and explainable insights to the user

At its core, the system answers the question:

Given a set of mission constraints, what is the most sustainable biological system and how will it evolve over time?

How We Built It

Deterministic Decision Engine

The system uses a rule-based scoring mechanism that evaluates:

  • Environment (Mars, Moon, ISS)
  • Mission duration
  • Resource constraints
  • Optimization goals

Each biological component is scored, and the system selects the highest-performing configuration. This ensures that results are consistent, explainable, and reproducible.

Simulation Engine

The simulation layer is fully deterministic and does not rely on AI.

  • Time progresses in weekly steps
  • Resource consumption and production are tracked
  • Risk evolves dynamically over time

A simplified conceptual model:

[ Risk(t) = Risk_{base} + \sum_{i=1}^{t} f(resources_i, system_stability_i) ]

Key systems include:

  • A water cycle with delayed recovery
  • An energy model based on production and consumption
  • Multi-factor life-support metrics such as oxygen and food supply

AI Integration

AI is used strictly as an explanation layer.

It is responsible for:

  • Summarizing system outputs
  • Explaining decisions and trade-offs
  • Suggesting improvements

AI does not influence core decision-making. This hybrid approach preserves reliability while improving interpretability.

System Architecture

  • Backend: FastAPI
  • Frontend: React with Vite
  • Database: PostgreSQL
  • Deployment: Vercel and Railway

Additional features include:

  • Rate limiting
  • Response caching
  • Fallback mechanisms when AI is unavailable

Challenges

Balancing AI and Determinism

Early attempts to use AI for decision-making resulted in inconsistent outputs. This led to a redesign where AI was restricted to post-analysis only.

Simulation Realism

Initial simulations were overly linear and unrealistic.

Improvements included:

  • Delayed resource recovery systems
  • Duration-aware risk scaling
  • Multi-factor dependencies between variables

User Interface Stability

Dynamic UI elements caused layout issues.

This was resolved by restructuring the layout and improving component hierarchy and state management.

Deployment Issues

Problems included routing errors, CORS configuration, and backend availability.

These were solved through proper environment configuration, routing fixes, and health check systems.

What We Learned

  • Deterministic systems provide strong reliability and clarity
  • AI is most effective as a supporting layer
  • Simulation design requires balancing realism and stability
  • Delivering a working system is more valuable than over-engineering

Future Work

  • Integration of more advanced biological models
  • Use of real-world datasets
  • Support for multi-mission optimization
  • More advanced AI-assisted analysis

Conclusion

AstroTEA is a working prototype of a decision intelligence system for space agriculture.

It combines deterministic decision-making, realistic simulation, and AI-based explanation to provide a stable and understandable solution for mission planning.

This project represents a step toward sustainable life-support systems for future space missions.

Built With

Share this project:

Updates