Autonomous object tracking with vision based control using a 2DOF robotic arm

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Real-time object tracking with robotic manipulators is crucial for applications like manufacturing and healthcare.
• Traditional systems face challenges in sensor surveillance, stability, and complexity.
• Existing visual servoing approaches have limitations that this study aims to overcome.

Purpose of the Study:

• To design a precise and responsive object-tracking system for robotic arms.
• To eliminate complexities associated with traditional tracking mechanisms.
• To develop an autonomous vision-based control system for moving object tracking.

Main Methods:

• An image-based visual servoing (IBVS) approach was employed for a 2-degree-of-freedom (DOF) robotic arm.
• A deep learning-based object detection framework was utilized for real-time object identification and localization.
• A vision-based control technique was designed, integrating the object detection system's real-time response.

Main Results:

• The proposed deep learning controller demonstrated high accuracy and rapid response times in visual servoing tasks.
• Simulations using CoppeliaSim and experimental validation with a 2-DOF robotic arm confirmed the strategy's effectiveness.
• The system successfully tracked moving objects autonomously.

Conclusions:

• The developed deep learning-based vision control strategy offers a robust solution for robotic arm object tracking.
• The approach simplifies traditional methods by reducing reliance on complex mechanisms and multiple sensors.
• Further exploration into data-driven learning techniques can enhance the control scheme's adaptability and robustness.