Shape Sensing of Flexible Robots Based on Deep Learning

In this paper, a deep learning method for shape sensing of continuum robots based on multi-core Fiber Bragg Grating (FBG) fiber is introduced. The proposed method, based on an Artificial Neural Network (ANN), differs from traditional approaches, where accurate shape reconstruction requires a tedious characterization of many characteristic parameters. A further limitation of traditional approaches is that they either require multiple fibers, whose location relative to the centerline must be precisely known (calibrated) or a single multi-core fiber whose position typically coincides with the neutral line. The proposed method addresses this limitation and thus allows shape sensing based on a single multi-core fiber placed off-center. This helps in miniaturizing and leaves the central channel available for other purposes. The proposed approach was compared to a recent state-of-the-art model-based shape sensing approach. A 2-DOF bench-top fluidics-driven catheter system was built to validate the proposed ANN. The proposed ANN-based shape sensing approach was evaluated on a 40 mm long steerable continuum robot in both 3D free-space and 2D constrained environments, yielding an average shape sensing error of 0.24 mm and 0.49 mm, respectively. With these results, the superiority of the proposed approach compared to the recent model-based shape sensing method was demonstrated.