Michael D. Wheeler
PARIS -- Shipbuilding plants have sought welders with top-notch skills for years. Now, thanks to photonics, a new breed of welder -- the robotic kind -- could rival humans in expertise and efficiency.
The French Welding Institute developed RoboKid, a robot with a vision system that is based on laser triangulation, a distance measurement technique involving lasers and photodetectors. A rangefinder mounted on the robot's welding torch emits laser light from several perspectives.
Depending on the vision system, Servo-Robot Inc. in Boucherville, Quebec, Canada, installs one of two types of semiconductor laser diodes: one emitting at 780 nm or another emitting in the 810- to 830-nm range. The laser light bounces off the metal part that the robot will weld. The time it takes the laser to reflect to the 3-D camera head determines the location of the part. Software interprets the measurements, which come in at a rate of 40 images per second -- enabling the robot to "see" what it welds and automatically adjust the laser torch so it remains at the center of each weld.
The "seeing" robots already have found a home in shipyards in Cherbourg, France; at Framatome, a manufacturer of nuclear reactors; and at several automotive assembly plants. Manufacturing Engineers at each site have fitted the robots with a PC, so they can calculate welding strategies for their particular application. The com- puter also provides camera control and image processing capabilities that aid production.
Servo-Robot Inc.'s 3-D laser camera has brought a new level of accuracy and adaptability to robotic are welding.
By incorporating vision in the robot, scientists have eliminated a number of constraints that have hampered its "blind" predecessors. Those robots performed along preset trajectories and required reprogramming any time a different task was required, which often led to lengthy and costly shutdowns while engineers made the necessary adjustments to the software.
Jean-Louis Bréat, a member of the French Welding Institute, said that using a vision system makes it possible for the robot to adjust to real-world conditions. He said it enables accurate calculations of volumes and pass conditions, while the quality and regularity of the molds surpass the results obtained in manual welding.