Adaptive Control of Welding Robot with FPGA

Abstract

Depending on the development of robotic technologies and systems, robots have begun to be preferred in the industry for jobs that are dangerous and tiring for humans. The most common are welding robots, which have application areas such as automotive, household appliances, defense industry, medical, and aerospace. Welding robots automate welding applications and provide speed, time, and safety to the manufacturer. They can sustain heavy-duty cycles in an unchanging and uninterrupted manner. However, since automating the welding process is not under human control, it creates an uncontrolled process against possible errors. Current welding robots are programmed by an operator and the welding process is performed depending on the parameters determined by the operator. In addition, failure to maintain precision in measurements during the weld path programming causes errors. During welding, robot systems are unresponsive to known problems such as material distortion and arc blowing. The fact that the parameters are preloaded into the system and the absence of a real-time control structure makes the welding robots uncontrolled against errors that may occur during welding. This situation causes work, time, productivity, and financial loss. Different quality measurement and evaluation systems have been proposed in the literature after welding processes performed with welding robots. However, welding quality measurement was provided in these systems, and an error-proof control system was not developed. However, there is a need for a real-time control mechanism for automatic determination of the weld path. In this thesis study, a real-time system is proposed to prevent errors before they occur, to automate the programming process, and to provide an adaptive control structure in welding works performed by welding robots. The proposed system has been implemented on Field Programmable Gate Arrays (FPGA) due to its advantages, such as parallel processing and instant response to multiple tasks due to the computational load required for image processing and instant control. The data received using a camera and laser module was processed using image processing techniques to determine the weld path. The obtained data was transferred to the adaptive fuzzy logic controller as input information. The developed controller structure has obtained an adaptive speed control and welding path tracking system for welding robots. The experimental studies concluded that the developed system could successfully determine the weld path geometry and follow the weld path on a prototype.