Laser ablation is widely used for precision material removal, but controlling both ablation depth and surface quality remains challenging. No signal indicates when the target depth is reached, and offline metrology is slow and interrupts processing.

This work presents a fully integrated laser micro-processing workstation equipped with simple online sensors and a compact chromatic confocal sensor for real-time monitoring of ablation depth and surface quality. The sensor suite includes an infrared photodiode, an acoustic microphone, a fiber-coupled spectrometer, and a camera, all operating synchronously during processing. The chromatic confocal sensor measures depth between processing passes, while the other sensors acquire data continuously during ablation.

Experiments on silicon demonstrate quantitative depth prediction from individual sensor channels. Each sensor independently supports regression models achieving R² up to 0.8 and mean absolute errors near 20 µm across a thinning range of 0 to 350 µm, from a small experimental dataset. Multimodal fusion of all sensor channels matches the accuracy of the best individual sensor while improving robustness to sensor degradation or loss.

The chromatic confocal sensor enables direct closed-loop depth control without interrupting the process. Datasets acquired in-situ can train sensor-based models that operate in real time during ablation, providing a practical path toward a fully autonomous laser processing workstation.