A CyberOctopus in fluids: Unraveling the mechanics of architected soft system

Tekinalp, Arman

Permalink

https://hdl.handle.net/2142/127461 Copy

Description

Title

A CyberOctopus in fluids: Unraveling the mechanics of architected soft system

Author(s)

Tekinalp, Arman

Issue Date

2024-11-19

Director of Research (if dissertation) or Advisor (if thesis)

Gazzola, Mattia

Doctoral Committee Chair(s)

Gazzola, Mattia

Committee Member(s)

• Mehta, Prashant
• Krishnan, Girish
• Tawfick, Sameh

Department of Study

Mechanical Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Cosserat rods
• Mechanical intelligence
• Computational mechanics
• soft robotics
• velocity-vorticity formulation
• immersed boundary method
• flow--structure interaction
• distributed computing

Abstract

Muscular hydrostats such as elephant trunks and octopus arms have captivated both biologists and engineers due to their exceptional dexterity and reconfigurability. These structures, composed only of muscle fibers and connective tissue without any rigid skeletal support, exhibit unparalleled versatility in their movements. Despite the widespread interest and technological implications surrounding these systems, extracting design and control principles from their intricate nature has proven to be a challenge. In this context, this thesis endeavors to employ a combination of simulations, theoretical analyses, and experiments to delve into the underlying mechanisms at play. By investigating factors such as heterogeneous and non-linear muscle fiber organization, topology, control strategies, and interactions with the surrounding flow, we aim to unravel the sophisticated mechanical programs that are intricately embedded within octopus arms. Towards this goal, my research activities have been organized into four main thrusts: (1) Dynamics of soft, heterogeneous structures immersed in viscous flows. Soft slender architectures, found in both natural and artificial systems, play a crucial role across various scales and settings, from the components of organisms such as muscles, tendons, and bones, to applications in soft robotics and metamaterial design. However, accurately resolving these structures using traditional finite element methods can be challenging due to their extreme aspect ratios. This can be resolved by leveraging the Cosserat rod theory, which captures the 3D dynamics of slender elastic objects through a computationally efficient 1D Lagrangian representation. Towards this goal we developed an open-source ecosystem called PyElastica, which provides a comprehensive platform for simulating, analyzing, designing, growing, and controlling dynamic structures composed of slender elastic rods. We showcase the utility of PyElastica in a range of multiphysical systems across scales and environments, from simulating and control of single filaments, modeling root growth, and designing artificial muscles to full-scale, bio-inspired, connected and controlled cyber-octopus arm and a magnetically actuated cilia carpet. This broad applicability spans disciplines such as bio-mechanics, robotics, control, metamaterials, and more. Furthermore, to accelerate scientific advancements across the aforementioned domains and enable new ones, such as simulating immersed heterogeneous soft bodies for bio-hybrid swimming and underwater manipulation of soft robots, we have combined the power of Cosserat rod theory and velocity-vorticity formulation of Navier-Stokes equations. This synergy is achieved through the penalty immersed boundary technique, enabling seamless and consistent two-way coupling between multiple soft, slender bodies and the surrounding viscous fluid. Extensive benchmarking has confirmed the accuracy, robustness, and versatility of our algorithm. The resulting elastohydrodynamic solver paves the way for innovation in various scientific fields, including the modeling of bio-hybrid swimming soft robots, realistic muscular actuators in underwater environments (e.g., an immersed octopus), and the exploration of immersed soft metamaterials composed of fibrous networks (e.g., magnetic cilia carpets). These advancements hold promising applications in microfluidics, material design, and the field of underwater soft robotics. (2) Topology, dynamics, and control of a muscle-architected soft arm. Aforementioned software are ideal instruments to provide valuable resources for investigating the intricate mechanics of organisms, particularly muscular hydrostats like octopus arms or elephant trunks. These structures lack bones entirely, granting them remarkable dexterity and adaptability. The key to their unparalleled control over countless degrees of freedom lies in the arrangement of muscle fibers, which effectively represents a sophisticated mechanical program. In this thrust, we combine medical imaging, biomechanical data, live behavioral experiments, and numerical simulations to develop a bio-realistic octopus arm model comprising ~200 continuous muscle groups. The level of detail in our model cannot be captured by single-rod models, which overlook the structural complexity, heterogeneity, and anisotropy of the arm. These simpler models reduce muscular activity to cumulative (and often arbitrary) torque or force functions applied to a single rod, creating a disconnect from the three-dimensionally organized actuators responsible for generating these loads. In contrast, our approach captures the mechanics and nonlinearities arising from actuator deformations, providing a more accurate representation than single-rod models. Through this endeavor, we aim to unravel the complexity of these structures. Our findings shed light on the underlying mechanisms of 3D arm motions, highlighting the role of storage, transport, and conversion of topological and geometric properties facilitated by simple muscle activation templates. By assembling these templates into higher-level control strategies, combined with the arm's compliance, we demonstrate their effectiveness in a range of object manipulation tasks. These tasks pose additional challenges, such as the need to align suckers properly for sensing and grasping. Overall, our work elucidates fundamental design and algorithmic principles relevant to muscular hydrostats, robotics, and dynamics. Furthermore, it significantly advances our ability to model muscular structures using medical imaging, potentially impacting healthcare-related applications. (3) 3D feedback control of muscle-architected soft arms. The preceding topological control strategy involves manually composing simple muscle activation templates combined with the arm's compliance to achieve complex 3D motions and tasks. This research thrust aims to integrate the mechanical intelligence of octopus arms with a 3D feedback controller to enable efficient, automated control of an octopus-inspired soft arm with ~200 continuous muscle groups. This efficiency is rooted in the arm's muscular connectivity, where simple contractions naturally combine to generate complex 3D deformations. We evaluate the controller's performance on various tasks, including grasping, reaching, and tracking. Additionally, we introduce a semi-autonomous control strategy for helical grasping of tall poles, inspired by octopuses. Finally, we demonstrate the scalability of our approach by extending the control system to an eight-arm octopus model with ~1,600 muscle groups, successfully performing a dynamic tracking benchmark. (4) Octopus arms chasing cues in viscous flows. This thrust focuses on integrating the modeling capabilities, feedback controller, and computational framework developed in previous chapters to simulate an octopus arm reaching prey that releases cues in viscous flows. First, we extend the elastohydrodynamic solver by incorporating the diffusion and advection of passive fields, such as prey cues. We then immerse the octopus arm model, developed in earlier chapter, into the flow using the immersed boundary technique. Next, we propose two sensing algorithms: one that localizes the prey based on the gradients of diffused cues, and another that combines gradient based search with rheotaxis strategies. Finally, we couple both sensing algorithms with the 3D feedback controller and evaluate their performance in a benchmark where the octopus arm detects cues and reaches the target in viscous flows. Overall, this thesis presents novel physical discoveries and versatile numerical algorithms packaged into scalable software. These contributions enable the accurate modeling of complex fiber-based structures in fluid environments, with wide-ranging applications in engineering, healthcare, and medicine.

Graduation Semester

2024-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/127461

Copyright and License Information

Copyright 2024 Arman Tekinalp

Owning Collections