As the complexity of robot application increases, there is an growing need of off-line programming system rather than on-line programming method as a robot teaching method. In off-line programming system, since a task is planned through the systematical use of object models, environment models, and task specification, the productivity of the planned task is higher than that in on-line programming method.
In this thesis, a CAD-based off-line programming system specifically for spot welding task in automobile industry is presented and implemented. The proposed system consists of the world modeled, the task planner, and the graphic simulator. As the world modeled, a data extraction method from the existing general CAD system, Auto CAD, is presented for reuse of the existing model data produced by a model designer. A collision free motion planning and an optimal sequence planning are considered in the task planner, and they are treated as one subject and then, solved through the modified Traveling Salesman Problem (TSP) algorithm. To verify the planned sequence, a user friendly graphic simulator is implemented.
A set of welding points of a car is used to show that the generated welding sequence is collision-free and near optimal, and the result showed that the proposed system could enhance the efficiency of a spot welding robot in automobile industry.