Football Table Robot

This project aims to develop a robot that can play table football against a human. It uses a camera as a sensor to detect the ball's and players' positions on the table. Based on the detected positions, the robot sends signals to motors, which control the movements of the robot players. You can find the full code here

1. Setup and Libraries

Technologies:
OpenCV (cv2): For image processing and ball/player detection.
NumPy (np): Used for matrix manipulations and data handling.
ROS (Robot Operating System): For communication between components (e.g., sending motor commands to move robot players).


            captura = cv2.VideoCapture(2)  # Camera setup

            data = np.zeros((muestras, 2))  # Data collection for ball/player positions

            jugadores = np.zeros([20,4])    # Matrix for player positions

2. Player and Ball Detection

The camera captures the playing field, converting images to the HSV color space to track the positions of the ball and players.

Ball Detection: Identified by a green mask.
Player Detection: Uses color masks to detect the position of different players.

Technologies: Kalman Filter predicts the ball's trajectory.


            kfObj = KalmanFilter()  # Initialize Kalman filter

            hsv = cv2.cvtColor(imagen, cv2.COLOR_BGR2HSV)  # Convert image to HSV

            verde_bajos = np.array([74,55,67], dtype=np.uint8)  # Ball color range

3. Control System: Motor Commands

Based on the detected positions of the ball and players, the system sends signals to control the robot’s motors through ROS publishers.

Publishing Motor Commands:
Goalkeeper (PubArquero)
Defenders (PubDefensas)
Midfielders (PubMedios)
Attackers (PubDelanteros)


            PubArquero = rospy.Publisher("/Arquero", Quaternion, queue_size=20)

            PubDefensas = rospy.Publisher("/Defensas", Twist, queue_size=20)

            PubMedios = rospy.Publisher("/Medios", Quaternion, queue_size=20)

            PubDelanteros = rospy.Publisher("/Delanteros", Quaternion, queue_size=20)

ROS Messaging: Quaternions and Twist messages are used to represent positions and movements.

4. Prediction and Player Movement

The Kalman filter predicts the future position of the ball, allowing the robot to react faster and more accurately. The detected positions of the players are used to move them to block or hit the ball.


            predictedCoords = kfObj.Estimate(x1, y1)

            cv2.circle(imagen, (predictedCoords[0], predictedCoords[1]), 10, [0,255,255], 2)

5. Visualization

The system visualizes the game by drawing rectangles around the detected players and circles around the ball, displaying both real-time and predicted positions.


            cv2.putText(imagen, "x=" + str(x1), (x1, y1 + 20), font, sletter, (255, 255, 255), 2, cv2.LINE_AA)

            cv2.imshow("original", imagen)

Football Table Robot: Camera-Controlled AI Opponent

1. Setup and Libraries

2. Player and Ball Detection

3. Control System: Motor Commands

4. Prediction and Player Movement

5. Visualization

Technologies Summary