System Setup and Components
This setup uses a Raspberry Pi as the central computer to integrate hardware and AI, enabling the training of a mouse to climb onto two wooden levers using reinforcement strategies.
1. Core Hardware And Wiring
The system is powered by a Raspberry Pi, which controls all the components through precise wiring (shown in Image 1). The wiring connects the tools and ensures seamless communication between the LLM and hardware.
2. Motorized Pulley System
The motorized pulley system (shown in Image 2) is responsible for lowering the water feeder as part of the positive reinforcement strategy. The pulley operates autonomously based on the LLM's decisions, providing water rewards to guide mouse behavior.
3. Indicator Lights And Feedback
The system includes indicator lights to reflect the activation of events (shown in Image 3):
Red Light: Left lever is triggered.
Green Light: Right lever is triggered.
Blue Light: Speaker is activated by the LLM.
These lights serve as real-time feedback for both debugging and policy refinement.
4. Full Cage Setup
The enclosure (shown in Image 4) serves as the experimental environment:
Two Wooden Levers: Positioned as the target for the mouse to climb.
Pulley System and Water Feeder: Positioned in the center for reinforcement.
Camera Feed: Tracks the mouse’s position in real time, sending data to the LLM.
Wiring and Lights: Strategically arranged around the enclosure for control and feedback.
5. Operational Flow
Real-Time Tracking: The camera feed, combined with the vision-based LLM, monitors the mouse's position and movement.
Stimulus Delivery:
LED Lights: Adjust lighting in specific zones of the cage.
Speakers: Emit auditory cues to influence the mouse’s behavior.
Food and Water Rewards: Delivered using the food dispenser and motorized pulley system.
Feedback Integration:
Indicator Lights: Show when levers are triggered or when the LLM decides to activate the speaker.
Real-time data is collected and fed back to the LLM for analysis.
Policy Refinement: The LLM generates, tests, and refines behavior-control policies to optimize success in guiding the mouse to climb onto the wooden levers.
LLM Interactions Diagram:
Summary
This setup combines AI-driven reasoning, real-time tracking, and precise hardware controls to train a mouse to climb onto two wooden levers. The system operates in a closed-loop where the LLM autonomously iterates and refines strategies.
Key components include:
Raspberry Pi as the control hub.
Vision-based LLM for real-time tracking.
Tools: LED lights, speakers, food dispenser, and motorized pulley system for reinforcement.
Feedback Mechanisms: Indicator lights and camera feed.
Last updated