Data from: Stretchable Arduinos embedded in soft robots
Data files
Aug 23, 2024 version files 202.66 MB
-
100_1.csv
-
100_2.csv
-
100_3.csv
-
100_4.csv
-
100_5.csv
-
100_ogain_1.csv
-
100_ogain_2.csv
-
100_ogain_3.csv
-
100_ogain_4.csv
-
100_ogain_5.csv
-
150_1.csv
-
150_2.csv
-
150_3.csv
-
150_4.csv
-
150_5.csv
-
150_ogain_1.csv
-
150_ogain_2.csv
-
150_ogain_3.csv
-
150_ogain_4.csv
-
150_ogain_5.csv
-
200_1.csv
-
200_2.csv
-
200_3.csv
-
200_4.csv
-
200_5.csv
-
DS10_1.csv
-
DS10_2.csv
-
DS10_3.csv
-
DS10_4.csv
-
DS10_5.csv
-
DS10_ogain_1.csv
-
DS10_ogain_2.csv
-
DS10_ogain_3.csv
-
DS10_ogain_4.csv
-
DS10_ogain_5.csv
-
Fig_2A_Compiled_400str_Samples.csv
-
Fig_2B_ogain_VHB_0.25_22ms_150str_1000cyc_S8.csv
-
Fig_2C_Cyclic_Data_2022-09-23_09.55.36.csv
-
Fig_4D_ProMini_Sample_1_atPTF.png
-
Fig_4D_ProMini_Sample_1_initial.png
-
Fig_4D_ProMini_Sample_2_atPTF.png
-
Fig_4D_ProMini_Sample_2_initial.png
-
Fig_4D_ProMini_Sample_3_atPTF.png
-
Fig_4D_ProMini_Sample_3_initial.png
-
Fig_4D_ProMini_Sample_4_atPTF.png
-
Fig_4D_ProMini_Sample_4_initial.png
-
Fig_4D_ProMini_Sample_5_atPTF.png
-
Fig_4D_ProMini_Sample_5_initial.png
-
Fig_4D_raw_data.csv
-
Fig_4F_Force_1_10_200_2_nohead.csv
-
Fig_5_Lilypad_final.JPG
-
Fig_5_Lilypad_initial.JPG
-
Fig_5_RGB_final.JPG
-
Fig_5_RGB_initial.JPG
-
Fig_5_SD_final.JPG
-
Fig_5_SD_initial.JPG
-
Fig_S1_raw_data_oGaIn.csv
-
Fig_S10_Compiled.csv
-
Fig_S2_test_2_amp_sweep_ogain_1500um_for_MATLAB_plus_egain_trial2.csv
-
Fig_S2_Test_3_flow_sweep_ogain_1500um_w_egain_trial_3.csv
-
Fig_S4_ogain_direct_comp_compiled.csv
-
Fig_S5_400str_compiled_0Ohm.csv
-
ogain_200_2_1.csv
-
ogain_200_2_2.csv
-
ogain_200_2_3.csv
-
ogain_200_2_4.csv
-
ogain_200_2_5.csv
-
README.md
-
VHB_ogain_1.csv
-
VHB_ogain_2.csv
-
VHB_ogain_3.csv
-
VHB_ogain_4.csv
-
VHB_ogain_5.csv
-
VHB_Proper_backing_1.csv
-
VHB_Proper_backing_2.csv
-
VHB_Proper_backing_3.csv
-
VHB_Proper_backing_4.csv
-
VHB_Proper_backing_5.csv
Abstract
To achieve real-world functionality, robots must have the ability to carry out decision-making computations. However, soft robots stretch and therefore need a solution other than rigid computers. Examples of embedding computing capacity into soft robots currently include appending rigid printed circuit boards (PCBs) to the robot, integrating soft logic gates, and exploiting material responses for material-embedded computation. Although promising, these approaches introduce limitations such as rigidity, tethers, or low logic gate density. The field of stretchable electronics has sought to solve these challenges, but a complete pipeline for direct integration of single-board computers, microcontrollers, and other complex circuitry into soft robots has remained elusive. We present a generalized method to translate any complex two-layer circuit into a soft, stretchable form. This enabled the creation of stretchable single-board microcontrollers (including Arduinos) and other commercial circuits (including Sparkfun circuits), without design simplifications. As demonstrations of the method’s utility, we embed highly stretchable (>300% strain) Arduino Pro Minis into the bodies of multiple soft robots. This makes use of otherwise inert structural material, fulfilling the promise of the stretchable electronics field to integrate state-of-the-art computational power into robust, stretchable systems during active use.
README: Stretchable Arduinos embedded in soft robots
https://doi.org/10.5061/dryad.80gb5mkxf
This contains both the CSVs of data in the paper and the MATLAB files I used to process those .csv files. It also contains raw images used in some figures where data was obtained.
General:
Each MATLAB script starts with the figure name it corresponds to (i.e., Fig 5c), and the script will ask for a .csv file upload. Choose the .csv with the corresponding figure name in the title. The script will output the data as plotted in the paper. The code is sectioned and commented on for easy understanding. NOTE: Fig 2D requires the 50 .csv files without a figure name in the file name to be in the same folder as the Fig 2D MATLAB script. These files do not have the Figure 2D indicator in the titles. Run the script with the 50 other files here in the same location, and you will get that plot.
In more detail:
- Fig*2A*_Compiled_400str_Samples.csv is the electromechanical data (resistance (Ohms), displacement (mm)) of the five samples needed for Fig_2A_Plotting_400str_Single_Trace_Samples.m to plot the data.
- Fig_2B_ogain_VHB_0.25_22ms_150str_1000cyc_S8.csv is the electromechanical data (resistance (Ohms), displacement (mm)) of the five samples needed for Fig_2B_Plotting_cyclic_Single_Trace_Samples.m to plot the data.
- Fig_2C_Cyclic Data 2022-09-23 09.55.36.csv is the electromechanical data (resistance (Ohms), displacement (mm)) for Fig_2C_Plotting_cyclic_Single_Trace_Samples.m to plot the data.
- Fig*2D_*tack_tests.m is paired with the below files, which plot the data:
100_1.csv through 100_5.csv contain tack testing data for Slacker 1 against steel with time (s), displacement (mm), and force (n) data.
150_1.csv through 150_5.csv contain tack testing data for Slacker 1.5 against steel with time (s), displacement (mm), and force (n) data.
200_1.csv through 200_5.csv contain tack testing data for Slacker 2 against steel with time (s), displacement (mm), and force (n) data.
DS10_1.csv through DS1*0*_5.csv contain tack testing data for DS10 against steel with time (s), displacement (mm), and force (n) data.
VHB_Proper_backing_1.csv through VHB_Proper_backing_5.csv contain tack testing data for Slacker 1 against steel with time (s), displacement (mm), and force (n) data.
100_ogain_1.csv through 100_ogain_5.csv contain tack testing data for Slacker 1 against ogain with time (s), displacement (mm), and force (n) data.
150_ogain_1.csv through 150_ogain_5.csv contain tack testing data for Slacker 1.5 against ogain with time (s), displacement (mm), and force (n) data.
ogain_200_1.csv through ogain_200_5.csv contain tack testing data for Slacker 2 against ogain with time (s), displacement (mm) and force (n) data.
DS10_ogain_1.csv through DS1*0*_ogain_5.csv contain tack testing data for DS10 against ogain with time (s), displacement (mm), and force (n) data.
VHB_ogain_1.csv through VHB_ogain_5.csv contain tack testing data for Slacker 1 against ogain with time (s), displacement (mm), and force (n) data.
- Fig_4D_raw_data.csv contains the tensile testing data (force (N), displacement (mm)) used to plot Fig_4D_Plotting_Pull2SCF.m. The accompanying Fig_4D[...].png files are image files used to calculate strains at failure.
- Fig_4E_Plotting_Cyc2SCF.m plots the cycle to failure data results.
- Fig_4F_Force_1_10_200_2_nohead.csv is the raw tensile data (force (N), disp. (mm)) used in Fig_4F_Plotting_Cyc_Force_4_1_10_200.m to plot tensile test data.
- Fig_5[...].png files image files used to calculate strains at failure.
- Fig_S1_raw_data_oGaIn.csv is the raw XRD data used in Fig_S1_XRD_plot.m to plot the XRD curve.
- Fig_S2_test_2_amp_sweep_ogain_1500um_for_MATLAB_plus_egain_trial2.csv is the rheometer data (stress, strain, etc.) used with Fig_S2_process_amp_sweep_4_plots.m to plot the rheometer data.
- Fig_S2_Test_3_flow_sweep_ogain_1500um_w_egain_trial_3.csv is the rheometer data (viscosity) used with Fig_S2_process_flow_sweep_1_plot.m to plot the rheometer data.
- Fig_S4_ogain_direct_comp_compiled.csv contains electrotechnical data (resistance (ohms), disp. (mm)) used to plot with Fig_S4_400str_direct_geometric_comparison.m.
- Fig_S5_400str_compiled_0Ohm contains electrotechnical data (resistance (ohms), disp. (mm)) used to plot with Fig_S5_Plotting_400str_Single_Trace_Samples.m.
- Fig_S6_transmission_line_results.m contains the data (resistance (ohms), distance (mm)) and plots it.
- Fig_S10_Compiled.csv contains the tensile data (force (N), displacement (mm)) for the compared materials, and Fig_S10_processing_tensile_data_and_modulus.m plots it.
Methods
Data was collected with MATLAB, Arduino, or Instron (materials testing) software, and processed in MATLAB. Relevant details are below. Please see the paper for more information.
OGaIn fabrication and characterization:
XRD testing was completed on a Rigaku SmartLab machine with a 2 mm window. SEM images were taken on a Hitachi CFE SU8230. EDS analysis was performed with a Bruker QUANTAX FlatQUAD (mapping at 5 kV and compositional analysis at 6 kV). Surface samples were prepared on 12.5 mm stubs on carbon tape. Cross-section samples were prepared on vertical sample holders on carbon tape, where a trace of OGaIn was frozen and then broken in half and mounted to the stub. Stretched cross sections were prepared by casting OGaIn on VHB, encapsulating it with rubber cement, and then stretching the substrate while applying it to the SEM stub. After mounting the sample, it was frozen and then broken to reveal the stretched cross-section.
The single-trace samples were created using the dimensions and electrode attachment techniques described in (45), with VHB tape and rubber cement as the substrate and encapsulant, respectively. The trace widths of OGaIn were 250 μm, with gauge length 25 mm (Fig. S3B, C). 0 Ω resistor samples were made using the same patterns and processes, but with a 0.25 mm gap in the middle of the trace, over which a 0402 0 Ω surface mount resistor was placed (Fig. S3E). The single trace samples were characterized using the custom setup previously described in (45), and illustrated in Fig. S3D. An Arduino-driven lead-screw actuator controlled the strain and strain rate, as resistance was recorded using a 4-point probe multimeter (BK
Precision).
Substrate viability characterization:
ASTM D6195-22 standard tack testing was carried out using an Instron 3345 tensile tester (maximum capture rate 2 ms, Fig. S9A). Five samples of each material were created as follows: VHB tape was adhered to the polyethylene terephthalate (PET) backing, ensuring no air bubbles were present. The DS10 (DragonSkin 10A, SmoothOn Inc.) samples were made by casting DS10 over the PET backing with a 0.5 mm drawbar. That same technique was used to create the samples of Slackers 1, 2, and 3. Slacker 1 used one part DS10 Part B, one part Slacker (SmoothOn, Inc.), and one part DS10 Part A (denoted, 100:100:100). Slacker 1.5 used the ratio 100:150:100, and Slacker 2 used 100:200:100. The same setup, sample creation, and procedure were used for the OGaIn adhesion as for the previously described tack testing, though instead of coming into contact with steel, the samples came into contact with molded OGaIn (Fig. S9B-E). A 25 ×25 mm, 1/16 in deep mold was created to fit onto the steel bar used for previous experiments (Fig. S9B). This mold was filled with OGaIn before each new sample, and a bar was drawn across the top surface before the removal of the mold (Fig. S9C).
Circuit characterization:
The five Arduino Pro Mini samples used for pull-to-failure tests were manufactured with a border on the edges such that fabric could be adhered, limiting the strain in those regions and ensuring that only the circuit board was straining. To set up, these fabric regions were gripped (Instron 2713-007) by the materials testing system (Instron 3345), and the wires for serial connection were plugged into an FTDI (Future Technology Devices International Limited) programming board (Sparkfun DEV-09716), which was connected to a laptop. A code that flashed four off-board LEDs and printed timestamps to the serial monitor was uploaded to the Soft Pro Mini. The laptop and Arduino were recorded by an external camera setup. The Instron captured force and displacement data while it pulled upward at a rate of 15 mm/min until serial disconnect occurred. The initial length and length at failure of the circuit were calculated from the camera footage (with one initial frame, and one frame just before serial disconnect), and from these, engineering strains were calculated. This same process was used to test the failure of the three additional example circuits, one sample each, though they were gripped with acrylic adhered to the circuit as the strain limiter. The Arduino Lilypad was strained until the LED stopped blinking at the rate specified in the script. The Sparkfun Sound Detector was strained with loud ambient noise to detect failure time. The Sparkfun RGB and Gesture Sensor sent color output through serial until an error reading serial occurred. For all measurements in this paper, we chose to calculate the strain clamp-to-clamp (field standard (17, 26–28, 36)). Cyclic testing of the soft Arduino Pro Minis used the same cyclic testing device as for the single-trace samples. The same additional grip areas were used, but instead of attaching fabric, laser-cut pieces of 1/8” acrylic were pressed onto either side of the excess VHB tape such that the four through holes fit onto the alignment pins of the cyclic tester (Fig. S11). The cyclic tester strained to 100% strain for 1000 cycles, while a time-lapse camera monitored the laptop screen, which concurrently displayed the serial output from the soft circuit, and the cycle number from the testing device. The number of cycles the circuit survived was recorded as the cycle before an error was seen in the serial output of the stretchable Arduino. After pull-to-failure and cyclic testing until serial disconnect, the samples were evaluated using the 2-point probe FLUKE multimeter to determine where the failure occurred (Fig. S7C). To determine the effect the circuitry had on the circuit substrate’s force-strain relationship, the same strain-limiting fabric pieces were attached to five stretchable Pro Mini circuits and five pieces of plain VHB tape. The materials testing system gripped these sections and strained each sample to 100% strain for 200 cycles, at 15 mm/min. Data from cycles 1, 10 (after Mullin’s effect (57), and 200, straining and relaxing, were isolated for each sample. The means and standard deviations were then calculated for the five samples, from cycles 1, 10, and 200, straining or relaxing.