Towards autonomous mobile robotic welding: Theoretical development and empirical validation

Expanding the use of robotics in industries that manufacture large products, such as ship-building, is a relatively new enterprise. In general, every ship is unique in its design and manufacture. Further, the size and scale of a typical ship combined with the high costs associated with dry-docks or real-estate immediately adjacent to the launch location has led toward a common manufacturing technique, in which the structural components of the ship are assembled in multiple locations with only the final assembly occurring in the most expensive location. Because of the unique issues in manufacture of these large scale systems, these environments will be called "unstructured" manufacturing environments. Advances in mobile robotics make them viable solutions for new methods for automation or mechanization in unstructured manufacturing environments. These systems hold the promise to mechanize and automate a number of specific manufacturing processes, such as welding. However, the research and development work to date tends to focus on component level offerings to this problem but not merged systems. For example, advanced work in welding systems adds intelligence and machine control to the electro-thermal process, but not the motion control process. Conversely, work in mobile robotic systems focus on kinematics, dynamics and control of mobility without consideration for interactions with a given manufacturing process. The research presented here will demonstrate a merger of mobile robotic and welding systems, beginning at the modeling level, and advancing through system calibration, validation, and implementation into unstructured manufacturing environments. Specifically, this research results in an integrated system model derived for a combination of the gas metal arc welding process, the weld bead geometry, and the mobile skid-steered robotic platform. Certain aspects of multiple weld models will be expanded to show the inherent coupling to the robotic welding system and describe how the models will be utilized to incorporate some of the dynamic parameters from the robot. A final model will be presented and empirically calibrated. The final model will be used to determine initial starting parameters for both the robot and welding systems based on user inputs. This model will be implemented on a mobile robotic system and will provide a weld setup tool to guide less experienced weld operators in the welding process. The system will define initial weld system parameters that can reduce tuning that is needed before the process is stable and performing properly. Also, this research will show the details of a process to validate and verify the MRWS as a viable commercial welding system.