AUTONOMY x Manufacturing

The English wheel shapes sheet metal into smooth, multi-curvature panels, but it depends on skilled craftsmen who adjust the wheel pressure and path by sight and feel. As part of the Sheet Smith testbed in the NSF HAMMER Engineering Research Center, we are building an autonomous English wheel in which a UR robot handles and wheels the sheet while sensors track how it deforms. An uncertainty-aware multi-view stereo system reconstructs the sheet during forming and estimates how far each measured point can be trusted (Chen et al., 2026). An adaptive digital twin, built on a Koopman operator with model predictive control, plans toolpaths toward a target shape and updates itself online as new deformation data arrive (Chen et al., 2026).

(in collaboration with Profs. Wei Chen, Jian Cao, and Qi Zhu)

2026

Uncertainty-Aware Multi-View Stereo for Geometry Reconstruction in Manufacturing

Zihan Chen , Guanzhong Hu , and Ping Guo

In Proceedings of the ASME 2026 21st International Manufacturing Science and Engineering Conference (MSEC2026) , Jun 2026

Abstract PDF

Multi-View Stereo (MVS) provides a flexible and low-cost solution for full-field three-dimensional (3D) geometry measurement in engineering and manufacturing applications. Compared with laser scanning or fringe projection profilometry, MVS offers greater scalability and measurement efficiency through direct imaging setups. As manufacturing system evolves toward intelligent and data-driven paradigms, the ability to quantify measurement uncertainty becomes essential for reliable process monitoring and control. However, despite extensive research on MVS reconstruction algorithms, rigorous uncertainty modeling and evaluation remain largely underexplored, limiting the credibility of MVS as a metrological tool. This study presents an uncertainty-aware framework for estimating the geometric uncertainty of 3D points reconstructed by MVS. The approach is based on the law of uncertainty propagation, incorporating both camera calibration uncertainty and image matching errors. A multi-camera imaging system is deployed to monitor the English Wheel metal forming process, which serves as a representative case involving non-Lambertian and low-texture surfaces. Camera parameter covariances obtained from experimental calibration and empirically characterized matching uncertainties are propagated through the multi-view triangulation relationships to estimate the uncertainty associated with each reconstructed 3D point. Building on this uncertainty estimation, an uncertainty-guided view selection strategy is proposed to identify stable camera combinations for reconstructing different surface regions. By excluding unreliable observations and favoring views with lower expected uncertainty, the method improves both computational efficiency and reconstruction reliability. Experimental results demonstrate that the proposed approach enhances reconstruction accuracy and robustness compared with conventional MVS methods that utilize all available views indiscriminately. The framework therefore enables more reliable real-time deformation monitoring and provides a foundation for uncertainty-aware 3D metrology in advanced manufacturing environments.

Adaptive digital twin of sheet metal forming via proper orthogonal decomposition-based Koopman operator with model predictive control

Yi-Ping Chen , Derick Suarez , Ying-Kuan Tsai , Vispi Karkaria , Guanzhong Hu , Zihan Chen , Ping Guo , Jian Cao , and Wei Chen

Journal of Manufacturing Systems, Jun 2026

Abstract PDF

Digital Twin (DT) technologies are transforming manufacturing by enabling real-time prediction, monitoring, and control of complex processes. Yet, applying DT to deformation-based metal forming remains challenging because of the strongly coupled spatial-temporal behavior and the nonlinear relationship between toolpath and material response. For instance, sheet-metal forming by the English wheel, a highly flexible but artisan-dependent process, still lacks digital counterparts that can autonomously plan and adapt forming strategies. This study presents an adaptive DT framework that integrates Proper Orthogonal Decomposition (POD) for physics-aware dimensionality reduction with a Koopman operator for representing nonlinear system in a linear lifted space for the real-time decision-making via model predictive control (MPC). To accommodate evolving process conditions or material states, an online Recursive Least Squares (RLS) algorithm is introduced to update the operator coefficients in real time, enabling continuous adaptation of the DT model as new deformation data become available. The proposed framework is experimentally demonstrated on a robotic English Wheel sheet metal forming system, where deformation fields are measured and modeled under varying toolpaths. Results show that the adaptive DT is capable of controlling the forming process to achieve the given target shape by effectively capturing non-stationary process behaviors. Beyond this case study, the proposed framework establishes a generalizable approach for interpretable, adaptive, and computationally-efficient DT of nonlinear manufacturing systems, bridging reduced-order physics representations with data-driven adaptability to support autonomous process control and optimization.

References

2026