Nexus relif

📖 The Problem In disaster scenarios (wildfires, floods), decision-making is the bottleneck. Human dispatchers are overwhelmed by data. Nexus Relief solves this by creating an Autonomous Hive Mind that manages a fleet of drones to extinguish fires and rescue survivors efficiently.

🛠️ Key Concepts Implemented This project demonstrates 3 Core Agent Concepts:

1. Multi-Agent System (Sequential & Hierarchical) The Commander (Brain): An LLM-powered agent that analyzes the entire map state. The Drones (Field Units): Autonomous units that execute specific movement/refuel tasks. Logic: The system uses a "Central Dispatch" architecture where the Commander assigns tasks based on proximity and fuel levels, resolving conflicts before they happen.
2. Tools & Function Calling The Agent does not just output text; it executes Python Functions to manipulate the simulation:

move_drone(id, x, y): Navigates the grid. refuel_drone(id): Triggered automatically when fuel is critical. Human Override Tool: A unique "Human-in-the-Loop" tool allowing the user to inject priority commands that override the AI's standard logic.

1. Sessions & Memory Short-Term Memory: The Agent receives the real-time "World State" (Fuel, Zones, Locations) at every turn. Long-Term Persistence: The Mission Log persists across Streamlit reruns, maintaining a history of tactical decisions and human orders. State Management: Usage of st.session_state to track the evolving simulation grid.

Built With

• python
• streamlit