Akademik Veri Yönetim Sistemi
5th International Conference on Emerging Smart Technologies and Applications, eSmarTA 2025, Ibb, Yemen, 5 - 06 Ağustos 2025, (Tam Metin Bildiri)
This paper presents a comprehensive review of error compensation techniques for robotic manipulators, addressing the critical need for enhanced precision in modern industrial applications. It establishes a systematic taxonomy of error sources that categorizes geometric and non-geometric inaccuracies based on their physical origins and temporal characteristics. The review critically analyzes three dominant compensation paradigms: physics-based modeling approaches that leverage rigorous kinematic formulations, data-driven machine learning techniques capable of capturing complex nonlinear error patterns, and hybrid methodologies that combine their complementary strengths. Through extensive comparative evaluation, it is demonstrated that hybrid physics-ML approaches achieve superior performance in dynamic industrial environments, with certain implementations reducing positioning errors by up to 97%. The analysis reveals fundamental trade-offs between theoretical precision, computational efficiency, and practical implementation across different methodologies. Emerging research directions, including digital twin-enabled adaptive compensation and edge-cloud collaborative architectures, are identified as promising solutions to current limitations. This review provides both a theoretical framework for understanding robotic error phenomena and practical guidelines for implementing compensation systems in industrial settings, serving as a valuable reference for researchers and practitioners advancing the state-of-the-art in robotic precision.