The welding station consists of an industrial welding robot, eye view camera, and a compact 3D weld pool surface sensing system (Ref. 40). The robot utilized in this study (depicted
in Fig. 4A) is a Universal Robot, UR-5,with six degrees of freedom (DOF).The UR-5 industrial robot is a sixjointed robot arm with a low weight of 18 kilos, lifting ability of 5 kilos, and
working radius of 85 cm, respectively.The repeatability of the robot is ±0.1 mm. The robot is connected to a controller,which is used to control the motions of the robot. There is a touch
pad user interface that allows the userto program, control, and move the robot.The robot can also be programmed using URScript, a script language developed by the robot manufacturer.
The robot (client) and PC(server) is communicated via Ethernet using TCP/IP protocol and socket programming.Figure 4B shows the developed compact 3D weld pool surface sensing system for our robotic welding system (Ref. 40). In this system, a low-power laser (19 by 19 structure light pattern) is projected to the weld pool surface,and its reflection from the specular
weld pool surface is intercepted and imaged by a CCD camera (camera 1 in Fig. 4B). It is known that arc light is an omnidirectional light source. Its intensity decreases quadratically with the distance traveled, but the laser, due to its coherent nature, does not significantly lose its intensity. Hence, it is possible to intercept the reflection of the illumination laser from the weld pool surface with an imaging plane placed at an appropriate distance from the arc. From the distorted reflection patternon the imaging plane and the assumption of a smooth weld pool surface, the 3D shape of the weld pool surface can be obtained. By using a specific image processing and reconstruction algorithm (Ref. 19), a 3D specular weld pool can be reconstructed in real time. This 3D weld pool geometry information will be utilized to correlate the welder’s movement and the 3D weld pool geometry to enable the adaptive and accurate control of the welding process. To verify the effectiveness and accuracy of the compact 3D weld pool surface sensing system,a spherical convex mirror with known geometry (Edmund Optics NT64-057) is used as a benchmark (Ref. 40). The reconstructed 3D weld pool surface is depicted in Fig. 4C. It is
observed that most of the height errors are within 0.06 mm, especially in the central range of the weld pool.