2D image to 3D CAD model mapping of a T-joint pipe mapping for an intuitive robot teaching system

Abstract

This report presents the mapping process for 2D image with 3D CAD model of cylinders and T-joint pipes. The mapping is the critical process to conduct the intuitive robot path teaching to complete the welding tasks. This project aims to improve the intuitive method for robot path teaching using augmented reality technology. The system setup and augmented reality environment was developed by SimTech.

Traditional robot path teaching methods that rely on teaching pendants are known to be tedious, time-consuming and unsafe. By fusing the sensor information from a camera, an augmented reality environment was created. With the aid of a virtual CAD model, it is able to reconstruct the 3D space visible on the camera image. Applying this technology in a human and robot interaction system have improved the teaching process efficiency.

The pipes are widely used in manufacturing and welding on cylinders’ intersection points are difficult for traditional robot. The pipe image and CAD model is used to carry out robot path teaching for cylinder intersection welding more intuitively. Unlike other shapes with feature points, the pipes that consist of two cylinders have no direct feature points. This report describes how the algorithm conducts feature points determination for cylinder and pipe mappings. This report also presents the homogenous transformation matrix calculated to construct the 3D T-joint Pipe CAD model using OpenGL. In the report, the implementation of the algorithm in the system is described using program flow chart, screenshot of system interface and mathematic calculation.

Lastly, the report summarizes the project findings and gives recommendations for future work.