Defense Date

2024

Document Type

Thesis

Degree Name

Master of Science

Department

Mechanical and Nuclear Engineering

First Advisor

Hong Zhao

Abstract

Soft robotics has drawn tremendous interest in recent years because the compliance and motion of soft robotics enable biocompatibility and versatility for many applications, such as human-machine interaction, wearable and assistive devices, and health monitoring. This study introduces a novel predictive modeling approach using neural networks for shape control of magnetic soft robots. The robots are made of silicone materials embedded with hard magnetic particles, which respond to the external magnetic field provided by a ring-type of permanent magnet. These robots, free from physical connections to external devices, i.e., non-tethered actuation, hold significant potential for applications in healthcare, such as artificial limbs and surgical aids. The research methodology combines image processing, inverse kinematics, and neural networks to analyze and predict the movements of the soft robot under magnetic field, aiming to optimize parameters that achieve desired robot shapes. A specialized setup allows for precise manipulation of a permanent magnet's position relative to the robot, facilitating controlled actuation. The implementation of neural networks is pivotal, providing a predictive model that can adaptively calculate the magnet position to achieve the targeted robot morphological changes. This capability is crucial for refining design and control strategies in real-time applications

Rights

© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission

8-12-2024

Available for download on Tuesday, August 12, 2025

Share

COinS