Enabling Automated Programming of Welding Robots by Intelligent Identification of Weld Joints from Engineering Models

Abstract—Robot systems for automated welding have become very common in high-volume industries like the automotive sector. There are, however, unsolved issues to be answered when adopting this manufacturing technology to environments with small batch sizes or even lot size one. The biggest obstacle is the fact that programming efforts for robot systems often exceed the time for manual welding. This paper hence discusses known approaches for facilitating and accelerating this task. One problem common to these approaches will be tackled by this paper: The automated identification of weld joints from any arbitrary engineering model. We therefore propose a process oriented approach for identifying candidate joints and ruling out those not feasible for automated welding. As a result of this approach a motion path of the welding robot can be derived automatically. I