![You are a professional scientific illustrator for top academic conferences (CVPR/ICRA). Please create a clear, high-resolution "Data Processing Pipeline Diagram" to illustrate the "episodic memory writing process" of an embodied intelligence system.
Layout and Flow: Use a linear horizontal layout from left to right, including five key stages, connected by arrows.
1. First Stage: Raw Data Acquisition
Icon: An icon of a terminal window or log file with code or text lines.
Label: Habitat-Lab Raw Logs.
Description: Represents the raw interaction data output from the simulator.
2. Second Stage: Data Cleaning
Icon: A funnel or gear icon.
Label: Data Cleaning & Parsing.
Details: Indicates the removal of invalid data and formatting.
3. Third Stage: Structured Encapsulation
Icon: A neat document with the words "(Obs, Action, Outcome)" faintly visible.
Label: Prompt Structuring.
Details: Represents the encapsulation of cleaned data into experience triplets.
4. Fourth Stage: Vector Encoding
Icon: A diagram illustrating the transformation of text into a numerical matrix (e.g., a document passing through a neural network node and becoming a grid of numbers [0.1, 0.8...]).
Label: Embedding Encoding.
Details: Represents the transformation of natural language into high-dimensional vectors.
5. Fifth Stage: Storage
Icon: A cylindrical database icon with a vector grid pattern.
Label: ChromaDB Vector Store.
Details: The end of the process.
Visual Style:
Style: Modern academic flat 2D vector style (Flat Vector Art).
Color Scheme: Maintain consistency, using "Academic Blue" as the primary color and "Tech Grey" as the secondary color. Arrow colors should be dark and clear.
Font: Use a clear sans-serif font (Arial/Helvetica style), black or dark gray.
Background: Pure white background.](/_next/image?url=https%3A%2F%2Fpub-8c0ddfa5c0454d40822bc9944fe6f303.r2.dev%2Fai-drawings%2FUfvBGKD0YGX5OcBXzY3IQSo5YA3phloh%2Fe459777b-47d2-496c-bed1-8bc917030392%2Fe3e97e66-ed0d-41cd-9ce2-25a3615dcd86.png&w=3840&q=75)
Prompt Description
You are a professional scientific illustrator for top academic conferences (CVPR/ICRA). Please create a clear, high-resolution "Data Processing Pipeline Diagram" to illustrate the "episodic memory writing process" of an embodied intelligence system. Layout and Flow: Use a linear horizontal layout from left to right, including five key stages, connected by arrows. 1. First Stage: Raw Data Acquisition Icon: An icon of a terminal window or log file with code or text lines. Label: Habitat-Lab Raw Logs. Description: Represents the raw interaction data output from the simulator. 2. Second Stage: Data Cleaning Icon: A funnel or gear icon. Label: Data Cleaning & Parsing. Details: Indicates the removal of invalid data and formatting. 3. Third Stage: Structured Encapsulation Icon: A neat document with the words "(Obs, Action, Outcome)" faintly visible. Label: Prompt Structuring. Details: Represents the encapsulation of cleaned data into experience triplets. 4. Fourth Stage: Vector Encoding Icon: A diagram illustrating the transformation of text into a numerical matrix (e.g., a document passing through a neural network node and becoming a grid of numbers [0.1, 0.8...]). Label: Embedding Encoding. Details: Represents the transformation of natural language into high-dimensional vectors. 5. Fifth Stage: Storage Icon: A cylindrical database icon with a vector grid pattern. Label: ChromaDB Vector Store. Details: The end of the process. Visual Style: Style: Modern academic flat 2D vector style (Flat Vector Art). Color Scheme: Maintain consistency, using "Academic Blue" as the primary color and "Tech Grey" as the secondary color. Arrow colors should be dark and clear. Font: Use a clear sans-serif font (Arial/Helvetica style), black or dark gray. Background: Pure white background.
![You are a professional scientific illustrator for top academic conferences (CVPR/ICRA). Please create a clear, high-resolution "Data Processing Pipeline Diagram" to illustrate the "episodic memory writing process" of an embodied intelligence system.
Layout and Flow: Employ a linear horizontal layout from left to right, comprising five key stages connected by arrows.
1. Stage 1: Raw Data Acquisition
Icon: An icon of a terminal window or log file with code or text lines.
Label: Habitat-Lab Raw Logs.
Description: Represents the raw interaction data output from the simulator.
2. Stage 2: Data Cleaning
Icon: A funnel or gear icon.
Label: Data Cleaning & Parsing.
Details: Indicates the removal of invalid data and formatting.
3. Stage 3: Structured Encapsulation
Icon: A neat document paper with the words "(Obs, Action, Outcome)" faintly visible.
Label: Prompt Structuring.
Details: Represents the encapsulation of cleaned data into experience triplets.
4. Stage 4: Vector Encoding
Icon: A schematic diagram of text being transformed into a numerical matrix (e.g., a document passing through a neural network node and becoming a grid of numbers [0.1, 0.8...]).
Label: Embedding Encoding.
Details: Represents the transformation of natural language into high-dimensional vectors.
5. Stage 5: Database Storage
Icon: A cylindrical database icon with a vector grid pattern.
Label: ChromaDB Vector Store.
Details: The endpoint of the process.
Visual Style:
Style: Modern academic flat 2D vector art style.
Color Scheme: Maintain consistency, using "Academic Blue" as the primary color and "Tech Grey" as the secondary color. Arrow colors should be dark and clear.
Font: Use a clear sans-serif font (Arial/Helvetica style), black or dark gray.
Background: Pure white background.](/_next/image?url=https%3A%2F%2Fpub-8c0ddfa5c0454d40822bc9944fe6f303.r2.dev%2Fai-drawings%2FUfvBGKD0YGX5OcBXzY3IQSo5YA3phloh%2Fdd103003-88b7-4b05-ba5c-3de35b8b102d%2F236b67f0-c53d-417c-af61-1da5617bb090.png&w=3840&q=75)
