FBG-Based Position Estimation of Highly Deformable Continuum Manipulators: Model-Dependent vs. Data-Driven Approaches

Conventional shape sensing techniques using Fiber Bragg Grating (FBG) involve\nfinding the curvature at discrete FBG active areas and integrating curvature\nover the length of the continuum dexterous manipulator (CDM) for tip position\nestimation (TPE). However, due to limited number of sensing locations and many\ngeometrical assumptions, these methods are prone to large error propagation\nespecially when the CDM undergoes large deflections. In this paper, we study\nthe complications of using the conventional TPE methods that are dependent on\nsensor model and propose a new data-driven method that overcomes these\nchallenges. The proposed method consists of a regression model that takes FBG\nwavelength raw data as input and directly estimates the CDM's tip position.\nThis model is pre-operatively (off-line) trained on position information from\noptical trackers/cameras (as the ground truth) and it intra-operatively\n(on-line) estimates CDM tip position using only the FBG wavelength data. The\nmethod's performance is evaluated on a CDM developed for orthopedic\napplications, and the results are compared to conventional model-dependent\nmethods during large deflection bendings. Mean absolute TPE error (and standard\ndeviation) of 1.52 (0.67) mm and 0.11 (0.1) mm with maximum absolute errors of\n3.63 mm and 0.62 mm for the conventional and the proposed data-driven\ntechniques were obtained, respectively. These results demonstrate a significant\nout-performance of the proposed data-driven approach versus the conventional\nestimation technique.\n