Battleship

Overview

Developed a complete robotic system that enables a NAO humanoid robot to autonomously play Battleship against a human opponent. The project was completed during my semester exchange at the Technical University of Munich (TUM), Germany.

The robot had to visually locate and understand the game board, physically place its ships using its arms, communicate with the human player through speech, and reason about previous moves to select subsequent attacks.

My Contributions

• Vision: Developed the board detection and localization pipeline using ArUco markers, enabling the robot to identify the physical game board and establish its coordinate frame.
• Control & Manipulation: Developed the robot arm control and manipulation required for the robot to physically place and interact with ships on the board.
• Voice Interaction: Implemented components of the voice-based interaction system, enabling the robot to communicate game actions and results with the human player.
• System Integration: Integrated the perception, control, manipulation, and voice components into the overall autonomous gameplay pipeline.

System

The system was organized into four main components:

• Vision: Detected and localized the Battleship board using an ArUco marker, providing a consistent reference frame for the robot.
• Manipulation: Planned and executed arm movements to physically place the robot’s ships onto the board.
• Voice Interaction: Enabled the robot to communicate with the human opponent, including announcing moves and responding to hits, misses, and sunk ships.
• Game Cognition: Maintained the game state, tracked previously attempted coordinates, reasoned about the opponent’s fleet, and selected subsequent moves.

The game was implemented on a 5×5 board with a fleet consisting of two 3-cell Destroyers and two 2-cell Submarines.

The overall system follows a: Vision → State Estimation → Decision Making → Control & Manipulation → Voice Interaction loop, allowing the robot to perceive and interact with the physical game environment while playing against a human opponent.

5x5 Board with an Aruco marker on top.

Flowchart depicting gameplay.

Technical Highlights

• Platform: NAO humanoid robot
• Perception: ArUco marker detection + board localization
• Calibration: Board-to-robot coordinate transformation
• Motion & Control: Arm motion control for physical interaction
• Manipulation: Autonomous ship placement on a physical board
• HRI: Voice-based interaction with human opponent
• Planning: Game-state tracking + autonomous move selection
• Integration: End-to-end perception, planning, control, manipulation, and HRI pipeline

Team

Sumit Patidar · Konstantinos Theodorou

References

[1] NAO humanoid. IEEE. Available at: https://robots.ieee.org/robots/nao/ (Accessed: 5 October 2022)

[2] Battleship. Wikipedia. Available at: https://en.wikipedia.org/wiki/Battleship_(game) (Accessed: 5 October 2022)