Astra: Advanced A* Multi-Agent Warehouse Pathfinding Optimizer

National Level Hackathon Submission • Google Collaboration

A high-fidelity 3D Digital Twin for autonomous robot logistics and battery management.
System Architecture • Installation

📌 The Problem

Modern automated warehouses lose an average of 5% to 20% operational efficiency due to two primary bottlenecks: Path-Clashing (multi-agent congestion) and Battery Downtime (suboptimal charging cycles)¹.
(Project Name) solves this by implementing a battery-aware Multi-Agent Pathfinding (MAPF) algorithm within a 3D environment, optimizing throughput while ensuring zero dead-battery incidents.

⚡ Key Features

We moved beyond simple pathfinding. Our robots are self-aware agents.

🧠 Intelligence	⚡ Power Management	🏗️ Environment
A* Algorithm Finds shortest paths dynamically avoiding static shelves.	Auto-Docking Robots self-charge at <20% battery.	3D Digital Twin Real-time visualization using Ursina Engine.
Collision Avoidance Real-time "traffic control" to prevent multi-agent crashes.	Priority Queuing Urgent tasks override charging (if safe).	Custom Layouts Load warehouse maps via .txt config files.

Comparision of Astra with Industry Benchmarks (Generated by Gemini CLI)

🧠 System Architecture

The Logic Flow

flowchart TD
    %% --- Custom Style Definitions to match the image ---
    %% Blue: Process nodes and Start
    classDef blueFill fill:#203864,stroke:#4472c4,stroke-width:2px,color:white;
    %% Green: Decision diamonds and Positive outcomes (Discount)
    classDef greenFill fill:#1e452a,stroke:#548235,stroke-width:2px,color:white;
    %% Red: Negative outcomes, Penalties, and End states
    classDef redFill fill:#631d1d,stroke:#c00000,stroke-width:2px,color:white;

    %% --- Main Pathfinding Loop ---
    Start([Start Pathfinding<br>Robot at Current Position, Target]):::blueFill --> Init[Initialize Open Set with Start Node<br>x, y, wait_count=0, cost=0]:::blueFill
    Init --> IsEmpty{Is Open Set Empty?}:::greenFill
    
    IsEmpty -- Yes --> Fail([Path Not Found Failure]):::redFill
    IsEmpty -- No --> Select[Select Node with Lowest F-Cost<br>Current Node]:::blueFill
    
    Select --> IsTarget{Is Current Node == Target?}:::greenFill
    IsTarget -- Yes --> Success([Path Found Success, Traceback Path]):::redFill
    IsTarget -- No --> Expand[Expand Neighbors & Evaluate]:::blueFill

    %% --- Neighbor Evaluation Subgraph ---
    subgraph NeighborEval [Neighbor Evaluation]
        direction TB
        
        Expand --> IsHighway{Is Neighbor a<br>Highway Lane?}:::greenFill
        
        %% Branch: Highway Logic
        IsHighway -- Y --> PrefDir{Moving in Preferred<br>Direction?}:::greenFill
        PrefDir -- Y --> HwyDiscount[Apply<br>HIGHWAY_DISCOUNT<br>0.3x Cost]:::greenFill
        PrefDir -- No --> HwyPenalty[Apply<br>HIGHWAY_WRONG_WAY_PENALTY<br>+50.0 Cost]:::redFill
        
        %% Branch: Turn Logic
        IsHighway -- N --> IsTurn{Is Movement a Turn?}:::greenFill
        IsTurn -- Yes --> TurnPenalty[Add<br>TURN_PENALTY<br>+3.0 Cost]:::redFill
        
        %% Branch: Blocked/Wait Logic
        IsTurn -- No --> IsTempBlocked{Is Neighbor Cell<br>Temporarily Blocked?}:::greenFill
        IsTempBlocked -- Yes --> ConsiderWait{Consider 'Stay Put'<br>Wait?}:::greenFill
        ConsiderWait -- Yes --> WaitPenalty[Apply WAIT_PENALTY<br>+1.1 Cost,<br>Increment wait_count]:::redFill
        ConsiderWait -- No --> BlockedCost[Standard Blocked Cell<br>High Cost]:::redFill
        
        %% Branch: Occupied Logic
        IsTempBlocked -- No --> IsOccupied{Is Neighbor<br>occupied by<br>another Robot Soft<br>Obstacle?}:::greenFill
        IsOccupied -- Yes --> SoftAvoid[Add<br>SOFT_AVOIDANCE Cost<br>+8.0 Cost]:::redFill
        
        %% Convergence Point: Calculate Costs
        HwyDiscount --> CalcTotal[Calculate Total<br>G-Cost & H-Cost]:::blueFill
        HwyPenalty --> CalcTotal
        TurnPenalty --> CalcTotal
        WaitPenalty --> CalcTotal
        BlockedCost --> CalcTotal
        SoftAvoid --> CalcTotal
        IsOccupied -- No --> CalcTotal
        
        %% Path Evaluation
        CalcTotal --> IsBetter{Is New Path<br>Better?<br>Lower G-Cost}:::greenFill
        
        IsBetter -- Yes --> Update[Update/Add Neighbor<br>to Open Set with<br>new Cost & Parent]:::blueFill
    end

    %% --- Feedback Loops ---
    %% These arrows go back to the main loop start
    IsBetter -- No --> IsEmpty
    Update --> IsEmpty

Loading

The Math: Battery-Weighted A*

Our heuristic function $f(n)$ incorporates a Power Penalty $P(b)$ to influence routing toward chargers when needed:
$$f(n) = g(n) + h(n) + P(b)$$ Where:

• $g(n)$: Distance from start node.
• $h(n)$: Estimated distance to target shelf.
• $P(b)$: Exponential penalty based on current battery charge level ($100 - b$).

📂 Project Structure

📂 AIPathFinder
├── 📂 core                     - Core algorithms (Pathfinding logic)
├── 📂 entities                 - Game objects (Robots, Shelves, Chargers)
├── 📂 models                   - 3D Assets (.obj/.glb models)
├── 📂 textures                 - Visual assets and skins
├── 📂 cloud_logs               - Telemetry logs for Google Cloud
├── 📄 main.py                  - Main Simulation Entry Point
├── 📄 mainUI.py                - User Interface & Menu System
├── 📄 warehouse_layout.txt     - Warehouse Grid Configuration
├── 📄 default_layout.txt       - Backup Layout
├── 📄 cloud_telemetry.json     - Real-time Data Sync
├── 📄 GOOGLE_CLOUD_Features.md - Cloud Documentation
└── 📄 requirements.txt         - Dependencies

🚀 Getting Started

Prerequisites

Python 3.10+,
Pip (Python Package Manager)

Installation

1. Clone the repository:
   git clone
https://github.com/shlok377/AIPathFinder.git

2. Create a Virtual Environment:
   python -m venv venv

3. Command Prompt
   .\venv\Scripts\activate.bat

4. PowerShell:
   .\venv\Scripts\Activate.ps1

5. Install dependencies:
   pip install -r requirements.txt

6. Run the simulation:
   python main.py
python mainUI.py

Warehouse Layout Map

X = Shelf, # = Charger, T = Truck, . = Empty Aisle

.#......#....#......#.
.T......T....T......T.
......................
......................
..XXXX.XXX..XXX.XXXX..
..XXXX.XXX..XXX.XXXX..
......................
......................
..XXXX.XXX..XXX.XXXX..
..XXXX.XXX..XXX.XXXX..
......................
......................
..XXXX.XXX..XXX.XXXX..
..XXXX.XXX..XXX.XXXX..
......................
......................
..XXXX.XXX..XXX.XXXX..
..XXXX.XXX..XXX.XXXX..
......................
......................
..XXXX.XXX..XXX.XXXX..
..XXXX.XXX..XXX.XXXX..
......................
......................

👥 The Team

Shlok Pathfinding Algorithm Video Editing 3D Modeling & Simulation

Jashn Camera Framework Cinematography Presentation Voiceover

Vishva UI/UX Integration Telemetry Presentation Voiceover

(back to top)