Dobot Magician Lite — Autonomous Tic-Tac-Toe Robot

RAS 545 · Midterm 1 · Arizona State University · Fall 2025

A fully autonomous Tic-Tac-Toe-playing robot: Minimax AI computes optimal moves, the Dobot Magician Lite physically draws symbols on paper, and a computer vision pipeline detects the human player's moves via camera.

---

Overview

The Dobot Magician Lite (Serial: DT-MGL-4R002-01E) is programmed to play a physical game of Tic-Tac-Toe against a human opponent. The robot draws the 3×3 grid, executes its moves by physically drawing 'X' or 'O' symbols with a pen, and uses the Minimax algorithm to guarantee optimal play. The human player draws their symbol on the board and the system tracks game state via keyboard input with a planned CV integration via opencv_to_gemini.py.

---

Repository Structure

.
├── main.py                  # Main game loop — orchestrates vision, AI, and robot control
├── minimax.py               # Minimax algorithm — optimal move selection
├── robot_draw.py            # Dobot drawing control — grid, X, O symbols
├── draw_grid.py             # Grid geometry and coordinate calculation
├── camera_det.py            # Camera detection and initialization
├── port_connec.py           # Serial port connection to Dobot
├── port_connection.py       # Improved serial port handler
├── Midterm_1_report.pdf     # 3-page project report (IEEE format)
└── README.md

---

System Architecture

Human input / Camera → Board state update
                              ↓
                     Minimax algorithm
                              ↓
                    Optimal move (i, j)
                              ↓
              Grid index → Dobot XY coordinates
                              ↓
                   Robot draws symbol on board

---

Technical Details

Game Logic

• Algorithm: Minimax with full game tree search — guarantees robot never loses
• Human symbol: 'X' or 'O' (player choice at start)
• Turn order: Human first or Robot first (player choice at start)
• Win detection: Checks all rows, columns, and diagonals after every move
• Draw detection: Full board with no winner → "D" result

Robot Drawing

Parameter	Value
Robot	Dobot Magician Lite
End effector	Pen / marker
Grid size	95.0 mm outer side
Z draw height	0.5 mm
Z travel height	20.0 mm
Repeatability	±0.2 mm
Max reach	340 mm
Software	Python + pydobot

Coordinate System

The operator jogs the pen to the top-left corner of the grid at startup. All subsequent cell coordinates are computed relative to this datum using grid cell size and padding offsets — no hardcoded absolute positions.

---

Setup & Usage

Requirements

pip install pydobot opencv-python

Run

python main.py

On startup:

1. Robot auto-detects serial port and connects
2. Operator jogs pen to top-left grid corner and confirms
3. Robot draws the 3×3 grid
4. Player selects symbol and who goes first
5. Game loop begins — alternating human and robot turns
6. Robot announces win / draw at game end

---

Lessons Learned

• Minimax guarantees optimal play but requires the internal board state to be accurate — human input errors bypass the strategy entirely
• Z-height calibration is critical: too high = no ink, too low = pen drag causes movement errors
• Defining a user-set coordinate datum at startup is far more robust than hardcoded absolute positions — adapts to any table setup
• Replacing CV board detection with keyboard input simplified deployment but removed the autonomous sensing capability

---

Course Info

• Course: RAS 545 — Robotic and Autonomous Systems
• Instructor: Prof. Mostafa Yourdkhani
• University: Arizona State University, Tempe AZ
• Semester: Fall 2025