Close-range real-time camera pose estimation and AR-guided alignment for large-scale industrial components

Precise alignment of large-scale industrial components on machine tools is essential to ensure machining accuracy, product quality, and process efficiency. Errors introduced during the setup phase can propagate throughout the manufacturing process, often resulting in costly rework. Conventional alignment methods rely on laser tracker systems, which, despite their high precision, require specialized equipment, skilled operators, and long setup times, making them expensive and operationally demanding. To overcome these limitations, this work presents a real-time collaborative camera pose estimation framework that simplifies and accelerates the alignment process. The proposed solution integrates predictive simulation, acquisition trajectory planning, and augmented reality (AR) to enable fast, accurate, and intuitive alignment, even for non-expert users. The system built upon the IDEKO VSET solution was further developed within the TACCO project, co-funded by EIT Manufacturing and the European Union. The framework starts from high-fidelity 3D models and employs a Monte Carlo-based optimization strategy to determine the optimal placement of auxiliary components, including calibrated scale bars, coded targets (igloos), and cross reference frame. Configurable image acquisition strategies allow users to balance accuracy and computational complexity. Candidate configurations are evaluated through a least-squares bundle block adjustment simulation, based on collinearity equations and a Structure from Motion approach, enabling uncertainty propagation analysis and the generation of 95% confidence error ellipsoids prior to data acquisition. A key innovation lies in the system’s ability to translate optimized planning solutions into real-time immersive AR guidance. Operators are guided in the placement of auxiliary components and camera positioning through an AR head-mounted display, ensuring complete and accurate data capture. The proposed approach achieves alignment accuracies closer to laser tracker systems while significantly reducing setup time, cost, and dependence on specialized personnel, offering a scalable and cost-effective alternative for industrial component alignment.