Intelligent Tracking of Mechanically Thrown Objects by Industrial Catching Robot for Automated In-Plant Logistics 4.0

Area of Science:

• Robotics and Automation
• Artificial Intelligence
• Manufacturing Systems Engineering

Background:

• Industry 4.0 smart manufacturing relies on automated systems, including robotic part transportation.
• Optimizing in-plant logistics requires accurate real-time tracking and prediction of thrown object trajectories for catching robots.
• Predicting the flight path of objects with varying shapes and properties presents a significant challenge.

Purpose of the Study:

• To develop a method for accurately predicting the 3D catching position of thrown spherical objects in real-time.
• To address the challenge of trajectory prediction in automated manufacturing logistics.
• To enhance the efficiency and productivity of in-plant transportation systems.

Main Methods:

• Development of a 3D simulated environment for controlled object throwing experiments.
• Utilizing multi-view geometry with simulated cameras to accurately observe object trajectories.
• Training an encoder-decoder bidirectional Long Short-Term Memory (LSTM) deep neural network on a large dataset of trajectories.

Main Results:

• The trained deep neural network achieved high accuracy in predicting the real-time trajectories of thrown spherical objects.
• The simulation environment allowed for precise experimentation with various object properties and initial conditions.
• The system demonstrated the potential for real-time interception by catching robots.

Conclusions:

• The proposed deep learning approach effectively predicts thrown object trajectories for automated manufacturing.
• This research contributes to advancing intelligent tracking and prediction capabilities in Industry 4.0 logistics.
• The method offers a viable solution for improving the accuracy and speed of robotic catching operations.