Search

The advance of minimally invasive interventional surgeries has accelerated the iterative upgrading of continuum surgical robots. High compatibility with robotic applications and high precision are key features of sensors for robot navigation. In this paper, a spun multi-core fibre (SMCF) 3D shape sensor for continuum surgical robots is proposed. The sensor is fabricated by ultraviolet (UV) light inscribing fibre Bragg gratings (FBGs) in the SMCF, showing good spectra characteristics. Shape reconstruction is realized by the Frenet–Serret (F–S) frame and sensing characteristics such as twisting, strain and temperature are systematically analyzed. Especially, the twisting characteristics of SMCF and MCF are compared and the feasibility of self-torsional compensation of SMCF is demonstrated. In addition, shape measurement experiments with three known shapes are performed. The mean relative error (MRE) for each shape is 0.49%, 1.90% and 5.13%, respectively. These results show that the proposed sensor can accurately measure 3D shapes with temperature and twisting compensation, enabling its application prospects in the online shape feedback of continuum surgical robots. [ABSTRACT FROM AUTHOR]

• This study presents an error analysis and correction method for multi-core fiber to improve the accuracy of shape sensing. • Monte Carlo method was used to simulate and analyze the sensing error. • An MLP neural network is built to reduce the sensing error, which is suitable for multi-core fiber with different structures. • The sensing experiments under different curvature confirmed that method improves the sensing accuracy of multi-core fiber. To meet the application requirements of accurate shape sensing for biomedical robotics and flexible morphing structure of aircraft etc, the error analysis and correction method for multi-core fiber is proposed. First, the relationship between the central wavelength shift and curvature of the multi-core fiber is analyzed which is based on the principle of multi-core fiber shape sensing, and a temperature decoupling method was researched. Considering the influence of the fiber structural parameters on curvature sensing, which is simulated by using the Monte Carlo method, and the simulation results show a strong influence among them. Next, to reduce the sensing errors introduced by the structural parameters, an error correction method based on regularized multilayer perceptron (MLP) networks is investigated, the proposed method does not require precise structural parameters of the multi-core fiber. Finally, some experiments with different curvatures are carried out, the results showed that the maximum root mean square error (RMSE) decreases from 2.03 m−1 to 0.007 m−1, and the minimum RMSE decreases from 0.01 m−1 to 0.001 m−1. The method this paper proposed has potential applications in shape sensing in the fields of biomedical engineering and aerospace. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel shape-sensing method based on deep learning with a multi-core optical fiber for the accurate shape-sensing of catheters and guidewires. Firstly, we designed a catheter with embedded multi-core fiber containing three sensing outer cores and one temperature compensation middle core. Then, we analyzed the relationship between the central wavelength shift, the curvature of the multi-core Fiber Bragg Grating (FBG), and temperature compensation methods to establish a Particle Swarm Optimization (PSO) BP neural network-based catheter shape sensing method. Finally, experiments were conducted in both constant and variable temperature environments to validate the method. The average and maximum distance errors of the PSO-BP neural network were 0.57 and 1.33 mm, respectively, under constant temperature conditions, and 0.36 and 0.96 mm, respectively, under variable temperature conditions. This well-sensed catheter shape demonstrates the effectiveness of the shape-sensing method proposed in this paper and its potential applications in real surgical catheters and guidewire. [ABSTRACT FROM AUTHOR]