Nivas Piduru

Nivas Piduru

Robotics student at Arizona State University. Building the future of autonomous systems.

Autonomous Maze Solver Using Computer Vision and AI

Overview

Developed a fully autonomous robotic system capable of solving rectangular mazes of any size (3×3, 4×4, 5×5) without human intervention. The system integrates computer vision for maze detection, BFS path planning for optimal solutions, and precise robotic control with millimeter-level accuracy.

Key Capabilities

  • Dynamic Maze Detection: Automatically detects maze size without manual configuration
  • Optimal Path Planning: BFS algorithm guarantees shortest path in 100% of cases
  • Precise Navigation: <2mm positioning accuracy with homography-based coordinate transformation
  • Zero Collisions: Perfect safety record across 20+ test runs

Technical Stack

  • Framework: ROS 2 (Humble) with modular three-node architecture
  • Computer Vision: OpenCV - perspective correction, grid extraction, color-based marker detection
  • Algorithm: Breadth-First Search (BFS) for guaranteed optimal paths
  • Coordinate Transformation: Planar homography using DLT and SVD
  • Hardware: Dobot Magician Lite robotic arm
  • Development: AI-assisted development with Claude AI via ROS MCP Server (40% faster development)

Results

  • 95% maze detection success rate
  • 100% optimal path finding (BFS guarantee)
  • ±3mm robot positioning accuracy
  • 8-12 seconds total solution time
  • Successfully tested on 3×3, 4×4, and 5×5 mazes

Project Duration: November 2025
Course: RAS 545 - Robotics and Autonomous Systems (Midterm 2)
Institution: Arizona State University