Integrated FBG Sensing and Adaptive Control for Surgical Continuum Robots With Simultaneous Shape and Force Decoupling

Accurate shape reconstruction and contact force estimation are critical for the safe operation of surgical continuum robots but remain challenging due to the inherent structural flexibility and signal coupling issues in sensors. This article proposes an integrated fiber-optic sensing framework that leverages machine learning algorithms to achieve high-precision decoupling of force and position information. Specifically, an Extreme Learning Machine (ELM)-based model is developed to map wavelength shifts to spatial coordinates, effectively mitigating the cumulative errors of traditional integration methods. To address the cross-sensitivity between large deformations and contact forces, a novel Seagull Optimization Algorithm-optimized ELM (SOA-ELM) is introduced for six-dimensional force decoupling. To validate the sensing performance, the proposed system is implemented in a closed-loop control framework combining fuzzy adaptive PID and adaptive impedance control. Experimental results demonstrate that the sensing system achieves a sub-millimeter shape reconstruction accuracy (Euclidean error reduced to 0.46 mm) and robust force tracking capabilities. This work provides a reliable signal processing solution for Fiber Bragg Grating (FBG) sensors in high-curvature applications, significantly enhancing the perception intelligence of minimally invasive surgical systems.