Development of an AI-Driven, Vision-Guided Spherical Manipulator for Automated Welding Operations
Supervisor Name
Ahmad Albalasie
Supervisor Email
abalasie@birzeit.edu
University
Birzeit University
Research field
Mechanical Engineering
Bio
Dr. Ahmad Albalasie is an Associate Professor in the Department of Mechanical and Mechatronics Engineering at Birzeit University in Palestine since 2016, where he also served as the Department Chair for a three-year term. He earned his Ph.D. in Mechatronics Engineering from Technische Universität Berlin, Germany, in 2016. Dr. Albalasie also holds a Master of Science degree in Automatic Control Technologies from Politecnico di Torino, Italy (2012), and a Bachelor of Engineering degree in Mechatronics Engineering from Palestine Polytechnic University (PPU), Palestine (2008). Dr. Albalasie began his academic career at PPU, where he served as a lecturer for one year and a research and teaching assistant for two years before that. His research interests are focused on variable stiffness actuators, robust control, optimal control (with a particular emphasis on model predictive control), haptic control, the control of under-actuated robots, and floating parallel marine robots. He has authored numerous publications in these areas, contributing significantly to the advancement of research in these fields.
Spherical robotic manipulators exhibit notable kinematic benefits, including superior dexterity and compact structural arrangements. Nevertheless, these systems are considerably underexploited in modern robotic applications. This research focuses on overcoming the limitations of traditional quality assurance in robotic arc welding by integrating an AI-driven vision system. The study emphasizes real-time defect detection, predictive process control, and adaptive optimization to ensure weld integrity within demanding industrial contexts. 1. Introduction and Problem Statement Current robotic applications reveal a marked inclination towards articulated multi-axis robotic arms, notwithstanding the evident advantages offered by spherical manipulator architectures. Despite the considerable operational latitude provided by articulated systems, they remain bound by essential limitations, including sophisticated mechanical arrangements and increased demands for computational governance. 2. Research Objectives This study endeavors to enhance paradigms of automated quality assurance through the methodical development of an AI-driven, vision-augmented robotic welding arm. The methodological framework of this analysis consolidates core tenets of real-time machine vision integrated with AI and e control algorithms to develop an intelligent monitoring protocol in which the integrity of the weld is continuously verified and process parameters are autonomously optimized within operational framework. 3. Methodology and System Architecture 3.1 Vision System Integration The proposed system incorporates a stereo vision configuration to facilitate ongoing process monitoring capabilities. This implementation allows for accurate tracking of weld pool geometry and seam joint localization throughout operational sequences, providing critical visual data for quality assurance. 3.2 AI-Driven Quality Assurance Quality verification is enabled via deep learning models, which support real-time detection. The visual data is processed to identify welding anomalies. This analytical approach enables continuous quality monitoring and facilitates immediate corrective actions during the welding process. 3.3 System integration and process control The system architecture employs the Robot Operating System 2 (ROS 2) framework in conjunction with secure communication protocols to facilitate data integration across system components. This configuration enables the AI models to analyze sensor data and automatically adjust welding parameters, while maintaining comprehensive process documentation for quality tracking. 4. Applications and Implications The developed platform demonstrates significant applicability in operational environments characterized by hazards conditions, stringent quality requirements, or the need for fully automated, zero-defect manufacturing processes. 5. Conclusion This research contributes to the progression of intelligent robotic welding technology through the amalgamation of AI-driven vision systems and real-quality assurance protocols, thereby addressing the prevailing challenges of defect prevention and post-process inspection in contemporary manufacturing applications.
