Digitizing manual assembly and operator training in industry 4.0: the role of inside-out tracking in augmented reality systems

The evolution of augmented reality (AR) technology is reshaping industrial engineering, particularly in manual assembly training. This dissertation explores the utilization of AR, focusing on inside-out tracking, as an innovative approach to digitizing and assessing manual assembly tasks. Inside-out tracking leverages sensors embedded within AR devices, such as cameras, accelerometers, and gyroscopes, to capture spatial information, which is then used to digitize the operator's actions and interactions. This approach contrasts with traditional outside-in tracking methods that require extensive external infrastructure, offering a more adaptable and scalable solution for training environments. The research develops a comprehensive framework for integrating AR technology into the digitization process within industrial settings. This framework captures, analyzes, and provides feedback on cognitive and motor skills critical to manual assembly. A comparative analysis of inside-out and outside-in tracking techniques is conducted to evaluate their effectiveness across various industrial scenarios, offering valuable insights into the optimal application of AR technology in different contexts. Central to the framework are the components of skill digitization, skill comparison, and feedback provision, all refined and enhanced through the integration of AR technology. The study employs advanced methodologies, including computer vision and machine learning, to ensure the accuracy and actionability of the digitized data. This allows for real-time, precise feedback that can significantly improve training outcomes. In addressing the practical challenges of implementing AR in industrial training, such as tracking accuracy, environmental adaptability, and interface design, the dissertation provides guidelines to facilitate the effective deployment of AR technology. These contributions align with the broader objectives of Industry 4.0, promoting the creation of more immersive, responsive, and scalable training systems. By enhancing the digitization process through AR technology, this research sets new standards in industrial training. It addresses existing limitations in manual assembly training and lays the groundwork for future innovations, contributing to the ongoing digital transformation of industrial engineering.

增强现实（AR）技术的演进正重塑工业工程领域，尤其在手工装配培训方面。本学位论文探讨AR的应用——以由内向外追踪（inside-out tracking）为核心——作为手工装配任务数字化与评估的创新方法。由内向外追踪利用AR设备内置的传感器（如摄像头、加速度计和陀螺仪）捕获空间信息，进而用于数字化操作员的动作与交互过程。该方法与传统的由外向内追踪（outside-in tracking）方法形成对比——后者需依赖大量外部基础设施——为培训环境提供了适应性更强、可扩展性更高的解决方案。本研究构建了一套综合框架，用于将AR技术整合至工业场景下的数字化流程中。该框架可捕获、分析手工装配所需的关键认知与运动技能，并提供反馈。研究对由内向外与由外向内追踪技术进行了对比分析，以评估二者在各类工业场景中的有效性，并为AR技术在不同情境下的最优应用提供了宝贵见解。框架的核心组件包括技能数字化、技能对比与反馈提供——所有这些均通过AR技术的整合得以优化与强化。本研究采用计算机视觉与机器学习等先进方法，确保数字化数据的准确性与可操作性。这使得实时、精准的反馈成为可能，从而显著提升培训效果。在应对工业培训中部署AR所面临的实际挑战（如追踪精度、环境适应性与界面设计）方面，本学位论文提供了指导方针，以促进AR技术的有效部署。这些研究成果契合工业4.0的宏观目标，推动构建更具沉浸式、响应式且可扩展的培训系统。通过AR技术增强数字化流程，本研究为工业培训树立了新标准。它解决了手工装配培训中的现有局限，为未来创新奠定了基础，助力工业工程领域的持续数字化转型。