Dynamic behaviors and embodied control principles in soft biological systems

Zhang, Xiaotian

Permalink

https://hdl.handle.net/2142/115935 Copy

Description

Title

Dynamic behaviors and embodied control principles in soft biological systems

Author(s)

Zhang, Xiaotian

Issue Date

2022-07-15

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

Gazzola, Mattia

Doctoral Committee Chair(s)

Gazzola, Mattia

Committee Member(s)

• Saif, Taher
• Bashir, Rashid
• Krishnan, Girish
• Kong, Hyunjoon

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Soft mechanics
• Computational mechanics
• Bio-physics
• Soft robotics
• Bio-hybrid robots

Abstract

Musculoskeletal systems consist of bones, muscles, tendons, ligaments and other connective tissues that altogether provide function and structure to natural creatures. One of the most intriguing aspects of these architectures is the often inseparable nexus between actuation and control, topology and mechanics, due to the continuum, non-linear nature of their constitutive elements. As a consequence, and in stark contrast with rigid-body robots, soft creatures can harness a wide range of deformations and structural instabilities to effectively cope with complex, unstructured and dynamic environments [1]. Thus, biological musculoskeletal architectures, due to their intrinsic distributedness, softness and compliance, exhibit the ability to outsource control tasks to their embodiments, an emerging paradigm denoted as embodied or mechanical intelligence [1]-[3]. In this context, the overarching goal of this thesis is to establish modeling, simulation and experimental approaches to investigate principles of embodied intelligence. Towards this vision, my research activities are organized into three main thrusts: (1) Establishment of a computational framework to model and simulate complex musculoskeletal architectures. Central to this framework is Cosserat rods theory. This theory captures 3D deformations and dynamics of elastic slender body objects via an inexpensive one-dimensional representation. These slender elements can then be organized and connected to synthesize active, heterogenous biological layouts. The utility of such approach is demonstrated through the simulation and analysis of a range of biological systems across scales and environments: from simple human joints to full-scale aquatic, terrestrial and aerial creatures. Moreover, this modeling approach is natively integrated with evolutionary optimization strategies, enabling in-silico inverse design of soft (biological) robots. (2) Investigation of embodied control principles of natural soft creatures. This thrust focuses on how biological locomotory performance could originate from the interplay between soft mechanics, muscular actuation and environmental effects. To gain specific insights, snakes are used as proxy. Snakes indeed encapsulate a number of relevant features: they are soft, slender creatures that extensively make use of compliance and environmental effects (friction, contact) to simplify control and achieve locomotory adaptivity. Moreover, from a modeling perspective, they are ideally suited to our Cosserat rods approach, given their intrinsic slenderness. Further, environmental physics based on contact and friction can be easily included and inexpensively simulated, thus facilitating our computational investigations. Combining theory and simulations, my analysis reveals that a variety of seemingly complex, 3D locomotory behaviors emerge instead from simple stereotypical actuation patterns, that are passively modulated and adapted by soft body mechanics and the frictional environment. Insights into such embodied control strategies may inform engineering control principles of soft slithering machines in complex, heterogeneous environments. Finally, snakes represent an ideal platform to study the integration of simple neural control models, based on the framework of coupled-oscillators, due to the characteristic time-periodic, wave-like nature of their muscular activations. This integrative approach will facilitate the in-silico understanding and investigation of control strategies derived from biological neural infrastructures. (3) Design of bio-hybrid systems, as an in-vitro discovery platform for embodied intelligence. Here, our modeling approach is utilized for the design of bio-hybrid robots, in which living components (muscles and neuron cells) are combined with soft scaffolds to generate motion. By collaborating with experimentalists, a synergistic approach has been developed, leading to successful realizations of both swimming and walking bio-hybrid machines. These include a recent demonstration in which a computationally designed walking robot is experimentally demonstrated in combination with mirco-electronics, enabling wireless control. Further, to explore bio-inspired control opportunities, neural clusters have also been integrated in a muscular bio-hybrid swimmer, again computationally designed, bringing sensory and decision-making capabilities on-board. These endeavors demonstrate the potential of bio-hybrid systems as discovery platforms for investigating the interplay between bio-inspired actuation, sensing and control in soft machines.

Graduation Semester

2022-08

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Xiaotian Zhang

Owning Collections