Image-Based 2D PCD for Morphological Analysis of Tendrils-Like Structure
81
features which are worth studying [8] to further push technological advancements. They possess specific climbing mechanisms, like stem twining or specialized organs such as tendrils [9], that can be used for designing robotic artefacts
with peculiar morphological compliance and integrated intelligent behaviour to
negotiate with the complicated surrounding.
Tendrils are specialized, long, filiform, and sensitive organs derived from
reproductive structures. Mechanical stimuli trigger in the plant to curl around
a supporting structure and direct its growth in an anti-gravity fashion to search
for light [10].
According to Darwin [11], tendrils move following three key phases: circumnutation in which they search and approach a support, contact coiling where
they touch to confirm the ideal support and start coiling around it, and freecoiling in which they drag themselves towards the support by tightening the
coil. The dynamic process of morphological changes reflects tendrils’ sensing and
decision-making capacity [12] during each perversion growth stage as a function
of the external environmental stimuli. Noticeably, many botanists investigated
the related morphology of tendrils focusing on their origin and development [9],
developmental molecular genetics [13], flexibility property [14], and attachment
adaptation [15].
Tendrils have inspired the development of several technologies, like artificial
springs [16], safety crutches [17], and robots for space [18] and environmental
[19] exploration. Recently, Must et al. [20] proposed a new tendril-like soft robot
with reversible actuation based on the plant cell water-balance mechanism.
Even though, experimental and theoretical analysis have been already conducted to describe and understand the mechanisms behind the morphing of
tendrils [21], the current challenge in plant-inspired soft robotics remains the
correct description of the shape evolution in such natural structures and the
relationship between this evolution and the environmental triggering.
We previously approached this issue starting from the morphological analysis,
by proposing and applying a piece-wise 2D clothoid spiral-based method that
uses a semi-automatic image processing to extract and to represent tendril-like
curling structures [22]. However, our approach was highly dependent on human
tuned parameters and thus we did not achieve a full automatic optimal selection
of segments.
In this paper, we extend our image-based methodology by approximating
curling structures with Euler spirals (or clothoids) model in Fresnel Integral
with rational approximated form, to limit the number of free parameters and
the computational complexity. Euler spirals are already known to well suit for
resembling natural curling shapes having linear relations between the curvature
and the arc length [23,24], but their application for the morphological modeling
of real structures is very limited [25]. Here, we intend to apply such strategy
and extend our previous work [22] to achieve the full automatic selection of
the optimal number of segments and properly represent different configurations
of natural tendrils. Automatizing this phase would allow fastening and make
repeatable the process of long sequences analysis, during morphological studies
of continuum growing structures.
81
features which are worth studying [8] to further push technological advancements. They possess specific climbing mechanisms, like stem twining or specialized organs such as tendrils [9], that can be used for designing robotic artefacts
with peculiar morphological compliance and integrated intelligent behaviour to
negotiate with the complicated surrounding.
Tendrils are specialized, long, filiform, and sensitive organs derived from
reproductive structures. Mechanical stimuli trigger in the plant to curl around
a supporting structure and direct its growth in an anti-gravity fashion to search
for light [10].
According to Darwin [11], tendrils move following three key phases: circumnutation in which they search and approach a support, contact coiling where
they touch to confirm the ideal support and start coiling around it, and freecoiling in which they drag themselves towards the support by tightening the
coil. The dynamic process of morphological changes reflects tendrils’ sensing and
decision-making capacity [12] during each perversion growth stage as a function
of the external environmental stimuli. Noticeably, many botanists investigated
the related morphology of tendrils focusing on their origin and development [9],
developmental molecular genetics [13], flexibility property [14], and attachment
adaptation [15].
Tendrils have inspired the development of several technologies, like artificial
springs [16], safety crutches [17], and robots for space [18] and environmental
[19] exploration. Recently, Must et al. [20] proposed a new tendril-like soft robot
with reversible actuation based on the plant cell water-balance mechanism.
Even though, experimental and theoretical analysis have been already conducted to describe and understand the mechanisms behind the morphing of
tendrils [21], the current challenge in plant-inspired soft robotics remains the
correct description of the shape evolution in such natural structures and the
relationship between this evolution and the environmental triggering.
We previously approached this issue starting from the morphological analysis,
by proposing and applying a piece-wise 2D clothoid spiral-based method that
uses a semi-automatic image processing to extract and to represent tendril-like
curling structures [22]. However, our approach was highly dependent on human
tuned parameters and thus we did not achieve a full automatic optimal selection
of segments.
In this paper, we extend our image-based methodology by approximating
curling structures with Euler spirals (or clothoids) model in Fresnel Integral
with rational approximated form, to limit the number of free parameters and
the computational complexity. Euler spirals are already known to well suit for
resembling natural curling shapes having linear relations between the curvature
and the arc length [23,24], but their application for the morphological modeling
of real structures is very limited [25]. Here, we intend to apply such strategy
and extend our previous work [22] to achieve the full automatic selection of
the optimal number of segments and properly represent different configurations
of natural tendrils. Automatizing this phase would allow fastening and make
repeatable the process of long sequences analysis, during morphological studies
of continuum growing structures.