Morphological Characterization Analysis of Thin, Continuum Structures based on Non-constant Curvature Clothoid Spiral --- Taking Inspiration from Climbing Organ - Tendrils

Continuum structure, characterized by thin and slender geometries, is common to find in nature and appears to be an indispensable and essential feature of both biological and soft robotic systems. Owing to the continuous deformation of morphological configurations, such structures incorporate the common characteristics of natural systems and can provide innovative and intelligent solutions in a continuum mechanism with high compliance and adaptation when interacting with the external environment. Taking the morphological evolution of Passiflora caerulea tendrils upon mechano-response as the ideal representative of continuum structures, this thesis proposed the novel marker-less methodologies to reconstruct the shape of thin, continuum structures and to analyze tendrils’ curling morphologies and movement, which are based on vision and clothoid curve models in 2D and in 3D cases, respectively. To reconstruct the 2D morphology of tendrils, we proposed three model-based approaches: single-segment clothoid, piecewise clothoid with the prescribed number of segments, and piece-wise clothoid model with the optimal number of segments. These methods have been numerically validated on both the continuum of biological and artificial structures and also compared with constant curvature fitting results with a high average accuracy (R2 > 0.9). Based on the 2D work, we reconstructed the morphological evolution of tendrils using stereo vision with points-correspondence by Fréchet distance and describe them using 3D piecewise clothoid curve. Then, about the dynamic analysis of stimulated tendrils, we deduced the morphologies and stimulus-response relationship of tendrils from the variation and distribution of curvature, torsion and tip movement, which are consistent with the botanists’ perspectives on sensitivity and transduction. In addition, we developed a sorting algorithm to order unarranged points for the extraction of continuum structures' center-line in images. It is effective and adaptive in dealing with different cases, such as discontinuity, branches, and with/without intersection. In conclusion, the main contribution of this thesis is to provide a solution for shape reconstruction and representation of thin-continuum structures (both natural and artificial) in planar and in space. In particular, the expected impacts of the present work are twofold: • at the biological level, to provide an effective and practical tool to analyze tendril-like plants on behaviors and growth evolution in response to external stimuli. • at the engineering level, to demonstrate that clothoid spiral is reliable and accurate to model and describe slender, continuum morphologies both in 2D and in 3D, which may guide the design and control of the tendril-inspired biomimetic robot with variable stiffness along the body length.