Design,Modelling And Control Of A Novel Soft Humanoid Robot Hand System

Continued technical advancement in robotic systems has resulted in increasing applications in which robots and humans interact in close proximity,and in some cases,physical touch.In contrast to traditional robots,which are built of stiff and inflexible components,soft robots,which are predominantly made of extremely compliant and pliable materials,have intrinsically safe qualities.These robots are perfect for interacting with humans in a safe manner while still functioning in highly dynamic conditions.Soft robotics is a fast-growing discipline that makes use of biomimetic design ideas,innovative sensor and actuation concepts,and improved manufacturing processes.In this dissertation,we aimed to design a soft humanoid hand,focuses on the exploration of the parameter optimization effects on soft actuators,dynamic model construction,and controller design of the soft actuator,and presents a new type of soft humanoid hand design,modeling,and control.The dissertation contributions are summarized as follows:The dissertation first presented an investigation of the impacts of modifying the soft actuator size and cross-section shape on soft actuator flexibility and forces generated at various applied pressures.This investigation may be used to improve the dimension ratio and cross-section form of the soft actuator.Increasing some dimension parameters,such as the actuator diameter "?",actuator length "L",and the number of air chambers,leads to improved performance.Conversely,decreasing the actuator wall thickness "t" improves actuator performance.Although increasing the air chamber height "H" increases the soft actuator flexibility,it reduces the grasping force.After simulating nine various cross-sections for the soft actuator,simulation revealed that the ellipse 2S*S shape is the best cross-section shape for the soft actuator in terms of performance.We also investigated the various causes of the dimension change impact.As a result,our study produced a wealth of information that may be utilized to assist and speed the soft actuator design process in order to develop robust and flexible Pneumatic Networks bending actuators.The nonlinear behavior of the materials utilized,the intricate geometries they build,and the vast range of movements they produce make identifying and forecasting the behavior of soft actuators difficult.We demonstrated how to use neural network technology to explain the motion and force produced by the pneumatic network bending soft actuator at varied input pressures.To validate the findings,three independent neural network models for three different modelling modes were built and tested with different input data sets.First,there is the dimension model,which deals with variations in the shape and geometry of the soft actuator and their impact on its reaction at varied pressure inputs.The second model is the free force model,which replicates the motion of a soft actuator in free space without the presence of any external disturbances.Finally,the blocked force model may be used to mimic a real-world soft actuator subjected to an external force.The performance of the bending soft actuator was assessed using Abaqus/CAE software simulation models,which provided quantitative insights into the actuators’ behaviors as well as input data sets for training the neural network models.Traditional manufacturing processes,such as casting,which need several fabrication stages to produce soft pneumatic structures,are inefficient and limit the development of soft pneumatic actuators.This work provides soft pneumatic actuators and a novel soft humanoid hand that are directly 3D printed in one manufacturing step without the need for postprocessing or support materials using low-cost and open-source fused deposition modelling(FDM)3D printers that use a commercially available soft thermoplastic poly(TPU)known as e SUN TPU e Flex 3D printer filament.The 3D printed soft actuators and soft humanoid hand can withstand high air pressure applications of up to 0.5 MPa without leakage and create a high bending ratio.The direct 3D printing approach may simply save a lot of time and effort when printing any complex shape like our present novel humanoid hand.We discuss the creation and characterization of our new soft humanoid hand,first displaying five different soft humanoid hand designs and the one that was picked.We provide a new soft humanoid pneumatic hand comprised of nine soft bending actuators constructed together in one hand that can match the mobility of the human hand as closely as possible,and we confirm this capability using hand sign language configurations.We can model the soft humanoid hand using FEM data and apply it in the control process.All control activities were performed using an Arduino Mega board and a software control model.We demonstrate the effectiveness of the proposed control system by three different types of trajectory control are performed.The suggested control model,which included all relevant sensors,was utilized to control the position of the soft humanoid hand fingers in real-time,resulting in high response performance with low inaccuracy with the reference signal.