Description:

This project explores the potential for human-robot interaction (HRI) in board games using the ReactorX-200 Robot Arm (RX200) by Trossen Robotics. Specifically, we aimed to develop a method for the RX200 to play Connect-Four against a human player, where the robot is able to recognize the current board state and make the best move in the given situation.

Features:

• The RX200 provides the capability of five degrees of freedom within its radial workspace of approximately 1.1 meters in diameter.

• The physical structure of the robot presents an opportunity for HRI with board games, as it can precisely navigate and move physical pieces that are commonly used in board games.

• Incorporation of a transposition table to store early moves that would otherwise take minutes to hours to calculate with the MiniMax algorithm, resulting in an average time of less than a tenth of a second for the RX200 to calculate moves.

Technologies Used:

• Computer vision (OpenCV) to enable the RX200 to recognize the current board state.

• Petri nets, which are a modeling and simulation tool with a graphical representation that allows the visualization of data flow in a finite structure. Petri nets were used to design the complex decision algorithms required for the RX200 to play Connect-Four against a human player.

Result:

We successfully created a human-robot interactable game of Connect-Four. The RX200 is able to efficiently calculate moves with an average time of less than a tenth of a second, due to the drastic reduction in the total time it takes when searching for moves by incorporating a transposition table and MiniMax search algorithm.