Ball-Catching System Using Image Processing and an Omni-Directional Wheeled Mobile Robot.

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Developing autonomous systems for dynamic object interception is challenging.
• Accurate real-time tracking and prediction are crucial for robotic ball catching.

Purpose of the Study:

• To design and validate an integrated robotic system for precisely catching thrown balls.
• To enhance the accuracy of ball trajectory prediction and robot navigation.

Main Methods:

• An omni-directional mobile robot equipped with dynamic stereo vision and a static camera.
• Kalman filter with deep learning for noise reduction and state estimation (position, velocity).
• Feedback linearization and PID control for robot navigation.

Main Results:

• The system accurately tracked the thrown ball using stereo vision.
• Deep learning-enhanced Kalman filter reduced visual measurement noise.
• The robot navigated effectively to intercept the ball.
• Experimental validation confirmed precise ball catching capabilities.

Conclusions:

• The developed system demonstrates high precision in autonomously catching thrown balls.
• The integration of advanced vision, filtering, and control strategies is effective.
• This research contributes to advancements in autonomous robotic interception systems.