3D complex curve seam tracking using industrial robot based on CAD model and computer vision

3D complex curve seam tracking using industrial robots applied in fields that affect human health such as welding, gluing, spraying, painting, etc. However, at present, manual performance by workers or teaching manipulators by using teach-pendant to recognize the planning trajectories curves, especially for 3D complex seam curves which often has complex shape, direction, small width are still the two main approaches. Through these two implementation methods, productivity is low and product quality is not uniform. So this study will present a practical method to overcome the above disadvantages by using a low-cost camera combining an image processing algorithm using a 3D CAD model and a pre-defined 3D curve profile that needs to be attached to that model to extract the contour on the object. This data is then transmitted to the industrial manipulator to perform the task of tracing the planned 3D complex curve. By using this method, the small width 3D complex curve is easily tracked and the system is not complicated. Moreover, the profile to be tracked can be modified easily and quickly by non-expert users with basic knowledge about CAD drawing and image processing. Through the result of the experiments, small error system and fast image processing time about 1.8 mm and 0.5 s respectively, the system proved that it meets the requirement of the production line and it can replace the worker. The system is also easy to maintain and setup.