In this study, we utilize simple light-emitting diodes (LEDs) and photodetectors (PDs) combined with an intelligent shape decoding framework to enable 3D shape sensing of a self-contained flexible substrate. Finite element analysis (FEA) is leveraged to optimize the LED-PD layout and enrich ground-truth data from sparse to dense points for model training. The mapping from light intensities to overall sensor shape was achieved with an autoregression-based model that considers temporal continuity and spatial locality. The sensing framework was evaluated on an A5-sized flexible sensor prototype and a fish-shaped prototype, where sensing accuracy (RMSE = 0.27 mm) and repeatability (Δ light intensity < 0.31% over 1000 cycles) were tested underwater. We validate an affordable alternative to FBG sensors with high-order sensing outputs, where demonstrations are supplemented in the below videos.

Table S1.  Performance of soft pneumatic actuators

TOPOGRAPHY PREDICTION RESULTS

Supplementary Materials: Figures S1, S2, S3, Equation S1

This Supporting Information includes: the detailed dimensions of the MRE valves, the dimensions of the four different MRE geometries investigated, and the detailed setup and parameters of the PID controllers used. Corresponding author Email:  sihan.wang@eng.ox.ac.uk 

This Supporting Information includes: _Supplementary text describing Preliminary Status Classifer, segmentation methods, model training and validation details; two supplementary tables, two supplementary figures and one supplementary video._ Corresponding author(s) Email:  _ hoychan@cityu.edu.hk; xinyueli@cityu.edu.hk; wenjli@cityu.edu.hk  _

This Supporting Information includes information regarding the magnetic field of the actuator magnet, MR-LF-S (which has the same geometry as MR-LF and a soft compartment), and a table comparing MR-LF to other small-scale, flexible magnetic crawler robots.  Corresponding author email:   yong.kong@utah.edu

This supporting information includes supplemental figures, movies, additional results, and the key steps in the neural network training algorithm.   Corresponding author Email:   yyang60@jhu.edu (Y.Y.)     libome@tsinghua.edu (B.L.)

This Supporting Information includes: a comparison of the REAL (Robot Ear Accomplished by Laser) with a typical vibration measuring system (Laser Doppler Vibrometers, LDV), frequency response of various materials on REAL and real-time analysis of REAL audio neural network model. Xiaoping Hong  Email:    hongxp@sustech.edu.cn

A document by Lingxiang Jia. Click on the document to view its contents.

Video S1. Magnetic actuation of the metamaterial on a flat surface. 

* Corresponding author. E-mail: X.Z.: xczhou@szu.edu.cn; L.Z.: lizhang@mae.cuhk.edu.hk; B.W.: benwang@szu.edu.cn

Complex environments, such as those found in surgical and search-and-rescue applications, require soft devices to adapt to minimal space conditions without sacrificing the ability to complete dexterous tasks. Stacked Balloon Actuators (SBAs) are capable of large deformations despite folding nearly flat when deflated, making them ideal candidates for such applications. This paper presents the design, fabrication, modeling, and characterization of monolithic, inflatable, soft SBAs. Modeling is presented using analytical principles based on geometry, and then using conventional and real-time finite element methods. Both one and three degree-of-freedom (DoF) SBAs are fully characterized with regards to stroke, force, and workspace.  Finally, three representative demonstrations show the SBA's small-aperture navigation, bracing, and workspace-enhancing capabilities.

Figure S1. Chemical characterization of a VO2memristive device. Energy-dispersive spectroscopic (EDS) elemental line-scan of a VO2 device.

This Supporting information includes supporting fugures and videos for our paper.  The information's Fig. 16 is the high-quality version of the article's Supplementary Movie 1.Corresponding author(s) Email:   nakashima@robot.t.u-tokyo.ac.jp  

 This file includes: Figures S1 to S12Movies S1 to S10 Other Supplementary Materials for this manuscript include the following: Movies S1 to S10

This Supplementary Information includes: Section S1- Fabrication method                                        Section S2- Actuation method                                          Section S3- Analysis of sixth-DOF torque                     Section S4- Experiments                                                     Figures S1-S31                                                                         Supporting Table                                                                   References                                                                                Other supplementary materials for this manuscript include the following:Supporting SI Videos S1-S10 Corresponding author(s) Email:   gzlum@ntu.edu.sg  

Figure S1. a) The memristive synaptic behavior with an ideally symmetric and linear weight update ability (constant ΔG for identical pulses) but limited conductance levels (N =20). b) Test accuracy for 10,000 images in the MNIST dataset obtained during the training of memristive DBN as a function of the training epoch (CDth=64).