Physics-Guided Spatiotemporal State Space Modeling for Lookahead Molten Pool Segmentation in Laser Wire-Feed Welding

2026-06-22

AI Summary

This study addresses the challenge of real-time weld pool perception in laser wire-fed welding, where sensor and actuator delays hinder accurate monitoring. To overcome this, the authors propose a physics-guided spatiotemporal state-space network that fuses historical coaxial grayscale images, process parameters, and arc electrical signals to predict the semantic layout—including keyhole, wire, and molten pool—up to 500 milliseconds into the future. The method innovatively integrates physical priors with temporal state-space modeling, incorporating keyhole motion awareness, temporal consistency constraints, and signed distance function supervision to significantly enhance geometric accuracy and robustness. Evaluated on 43 welding sequences, the approach achieves a mean Intersection over Union (mIoU) of 74.63%, with ablation studies confirming the critical contributions of temporal modeling and motion perception.

Abstract

Real-time weld-pool perception is critical for closed-loop control in laser wire-feed welding, where sensing, computation, and actuator response introduce unavoidable delay. This paper presents a physics-guided spatiotemporal state space network for lookahead weld-pool segmentation. The model uses historical coaxial grayscale images, welding process parameters, and aligned wire-state electrical signals to predict the future semantic layout of three physically meaningful regions: keyhole, wire, and molten pool. It combines a visual encoder, process- and sensor-conditioned feature normalization, patch-level temporal state space modeling, horizon-conditioned latent prediction, dense future feature prediction, and a motion-aware mask decoder. Auxiliary signed-distance-function supervision, temporal consistency, feature distillation, and fine-grained keyhole losses further constrain the predicted geometry and local motion. Experiments on a 43-sequence laser welding dataset show that the proposed WeldMamba reaches 74.63\% mIoU at a 500 ms lookahead. Ablation studies further show that temporal history, patch-level state space modeling, and keyhole motion awareness are the main contributors to robust future segmentation.

Problem

• weld-pool segmentation
• lookahead prediction
• laser wire-feed welding
• real-time perception
• spatiotemporal modeling

Innovation

• state space modeling
• lookahead segmentation
• physics-guided learning
• temporal consistency
• motion-aware decoding