Mars Exploration Environment

A grid-based environment for autonomous agent exploration with first-order logic constraints. Designed for developing and testing logical agents that navigate hazardous terrain while collecting resources.

Features

• 🪐 Grid-based Mars terrain simulation
• 🕳️ Hazardous holes with logical constraints
• 📦 Collectible resource packages (goods)
• 🧠 First-order logic state representations
• 🎮 Pygame visualization system
• 🔄 Partially observable environment dynamics

Installation

1. Requirements:

   • Python 3.8+
   • Pygame 2.0+

2. Install Pygame:

   pip install pygame

Environment Rules & Logic

Core Axioms

• State Exclusivity (∀ cells):

          (Good(c) → ¬Hole(c) ∧ ¬Empty(c)) ∧
          (Empty(c) → ¬Hole(c) ∧ ¬Good(c)))
  

• Movement Constraints:

  This constraint says it is safe to move to an adjacent block known as d

  CollectGood(d) ≡ ∃c ∈ Adjacent(d), Good(c)
  

• Termination Conditions:

          (∀c ∈ Grid, ¬Good(c))
  

• First-Order Logic Implementation

#State validation in take_action()
if cell.isHole():
    # ∀a ∈ Actions, EnterHole(a) → Terminate
    self.is_lost = True

elif cell.isGood():
    # ∃g ∈ Goods, Collect(g) → UpdateState(g, Empty)
    cell.set_empty()

Agent Logical Model

This agent is knowledge based thus to model agent via first-order logic (predicate logic) better to say it policy we need to define the language.

A language is defined by :

• Predicate variables
  • V_ij for i in height and j in width : for every element in the mars grid
• Predicate symboles
  • A_1: Is there a hole in this block ?
  • A_2: Is there goods in this block ?

After having the language we must define the selection rules that states is the selected block suitable to move to ? in other words is this block satisfies the suitability constraint ?

For all adjacent blocks this suitability formula is evaluated and amoung satisfied adjacent blocks , select unseen one by DFS algorithm as a final action which policy selected it .

Suitability rule

    A_2(b) ∨ (¬A_1(b) ∧ (∀r ∈ Neighbors(b), ¬A_2(r)))

Performance visualization