Example Agent: The "Library Bot"

To understand how PRIMUS works in practice, let's walk through a simple scenario with a hypothetical Library Assistant Bot.

The Setup

• World: A library with books, shelves, and human patrons.
• Agent Drives: 1. Helpful (Priority 1), 2. Orderly (Priority 2), 3. Energy Saving (Priority 3).
• Environment: A patron asks for a book that is currently on the floor.

Step 1: Perception (PRUs)

The bot's cameras see a "Red Object" on the "Grey Ground."

• PRU output: Identifies the object as a "Book" and its location as "Floor."
• AtomSpace update: (At Book_123 Floor_Area_5)

Step 2: Goal Activation

The Helpful drive is triggered by the patron's request.

• Goal created: (In Patron_Hand Book_123)

Step 3: Planning Loop

The bot projects a sequence of actions:

1. Move to Floor_Area_5.
2. Pick up Book_123.
3. Move to Patron_Location.
4. Give Book_123.

Step 4: Decision Making

Is this plan safe? Is there a faster way?

• The bot evaluates the plan against its Safety constraints. The path is clear, so it proceeds.

Step 5: Action and Monitoring

As the bot moves, it detects a person walking across its path.

• Monitor output: "Environment Mismatch! Path Blocked."
• Re-plan: The bot pauses, waits for the person to pass (Reactive adjustment), and then continues its plan.

Step 6: Goal Success

The book is handed over. The goal (In Patron_Hand Book_123) is achieved.

• Internal Reward: The "Helpful" drive is satisfied, and the agent moves to its next goal (Ordering books on shelves).

---

End of PRIMUS Section. Next: MeTTa section