Vision-Based 6D Pose Analytics Solution for High-Precision Industrial Robot Pick-and-Place Applications.

Area of Science:

• Robotics and Automation
• Computer Vision
• Manufacturing Engineering

Background:

• High-precision 6D pose estimation is crucial for industrial robot arms in manufacturing.
• Current methods face challenges in real-world pick-and-place applications.
• Accurate object localization is essential for automated assembly lines.

Purpose of the Study:

• To develop and validate an analytics-based solution for 6D pose estimation for industrial robot arms.
• To enable a Staubli TX2-60L robot arm for precise pick-and-place of metal plates.
• To improve the accuracy and success rate of robotic manipulation tasks.

Main Methods:

• Utilized an Intel RealSense D435 RGB-D camera for data acquisition.
• Developed a LabVIEW software infrastructure with NI Vision for image processing.
• Implemented particle filtering, equalization, pattern matching, and angle-of-inclination analytics for pose determination.

Main Results:

• The proposed analytical solution demonstrated superior accuracy compared to hybrid methods (YOLO-v8 + PnP/RANSAC).
• Achieved position errors under 2 mm and orientation errors below 1°.
• Attained a perfect success rate in pick-and-place tasks across four distinct scenarios.

Conclusions:

• The analytics-based 6D pose estimation method offers high precision for industrial robotic applications.
• This solution enhances the reliability and efficiency of pick-and-place operations in manufacturing.
• The proposed method provides a robust and accurate alternative to existing techniques.