How nasa-jpl/rosa works
Main Components
a) Core ROSA Framework:
• Main ROSA class (from src/rosa/rosa.py) as the central component
• Prompts system (src/rosa/prompts.py) connected to the main class
b) Tools Layer:
• ROS1 tools (ros1.py)
• ROS2 tools (ros2.py)
• Calculation tools (calculation.py)
• System tools (system.py)
• Logging tools (log.py)
c) External Dependencies:
• LangChain framework integration
• ROS System (both ROS1 and ROS2)
• LLM Interface
d) Demo Implementation:
• TurtleAgent as a separate component showing implementation example
Specific Design Elements
Use the following visual elements:
• Rectangle with rounded corners for the ROSA Core
• Hexagons for the Tools Layer components
Important Architectural Patterns to Highlight
• Plugin Architecture: Show how tools can be added/extended
• Bridge Pattern: Demonstrate ROS1/ROS2 compatibility
• Command Pattern: Illustrate natural language to action flow
Additional Details
Elements:
• Version compatibility badges (ROS1 Noetic, ROS2 Humble/Iron/Jazzy)
• Data flow indicators showing natural language input → processing → robot actions
• Extension points for custom implementations
• Testing framework integration
• Input Layer (Natural Language Processing)
• Core Processing Layer (ROSA Framework)
• Tools and Integration Layer
• Output Layer (Robot Control)