Skip to main content

Supported Features of the Suite

  • Visual Capability: Utilizing OpenCV for image processing and machine vision analysis, the suite can perform real-time detection and recognition of the chessboard and chess pieces' positions, ensuring accurate perception of the game state. Through real-time video stream input, the module can capture changes in the game state instantly, providing the latest information for subsequent decision-making.

  • Decision-Making Ability: Using a DQN network trained on PyTorch, the module can analyze the current game state, predict and generate the best chess-playing strategies to maximize the likelihood of winning. Through deep reinforcement learning techniques, the module can learn from a large amount of game data and continuously optimize its decision-making ability, gradually improving its level of play. With efficient algorithm design and optimization, the module can perform game analysis and decision-making in a short time, achieving fast and accurate moves.

  • Manipulation Capability: Using the MyCobot 280 robotic arm, the module can accurately locate the target position of chess pieces and place them in the specified positions through precise movement operations.

Currently supported models​

  • MyCobot M5, Pi
  • MyArm

Planned Features​

❔ Customizable wait time for gameplay. ❔ Limitation of gameplay attempts. ❔ Temporary retention of algorithm result display box. ❔ Selection of gameplay models. ❔ Robot debugging switch functionality. ❔ Debugging of movement path points.