Data from: Printing powerful soft robots on a desktop
Data files
Oct 30, 2026 version files 7.68 MB
-
Data_file_S1.xlsx
393.58 KB
-
Data_file_S2.zip
7.27 MB
-
README.md
15.43 KB
Abstract
This dataset accompanies the research article “Printing Powerful Soft Robots on a Desktop,” which presents high-power pneumatic oscillators and monolithically 3D-printed soft running robots that operate without electronic control. The double-expansion pneumatic oscillator enables a hexapod robot to reach a maximum running speed of 4.72 m/s, making it the fastest soft robot reported to date.
The dataset consists of two files. Data file S1 is an Excel spreadsheet containing tabulated experimental data underlying the figures in the article, including measurements of pneumatic and mechanical performance and robot locomotion. Data file S2 is a ZIP archive containing four 3D CAD designs: a single-expansion oscillator, a double-expansion oscillator, a single-expansion running robot, and a double-expansion running robot. These data and design files support the reproduction of the experimental results, fabrication of the devices, and further development of high-power pneumatic soft robotic systems. The dataset does not involve human participants or personal information.
Dataset DOI: 10.5061/dryad.sxksn03m0
Description of the data and file structure
The data were collected to characterize high-power pneumatic oscillators and monolithically 3D-printed soft running robots operating without electronic control. The experiments investigated the pressure, airflow, oscillation frequency, mechanical power, and locomotion performance of single-expansion and double-expansion pneumatic systems. Running robots were tested under different supply pressures and operating conditions, including indoor, outdoor, and aquatic environments. The dataset also includes 3D CAD designs of the oscillators and running robots to support fabrication and reproduction of the experimental results.
Files and variables
File: Data_file_S1.xlsx
Description: An Excel workbook containing tabulated experimental data underlying the figures in the associated research article. The data include measurements and calculated results related to pneumatic pressure, airflow, oscillation frequency, force, displacement, mechanical power, and running robot performance. Individual worksheets correspond to different figures or experimental measurements. Variable names and units are provided in the column headers of each worksheet. Blank cells are used where a measurement or value is not applicable or where data series have different lengths.
Worksheets and variables:
The workbook contains 30 worksheets, each named for the corresponding figure or supplementary figure in the article. Each description below identifies the experiment or analysis and defines the columns used in that worksheet. Repeated column groups represent separate measurements, configurations, or time series. Blank cells separate data groups or indicate that one series has fewer observations than another. Unless otherwise specified, pressures are in kilopascals (kPa), time is in seconds (s), frequency in hertz (Hz), speed in meters per second (m/s), force in newtons (N), displacement in millimeters (mm), and power in watts (W).
Fig. 2E — Bistable comparator force–displacement measurement.
Force versus displacement during loading and unloading of the comparator's bistable structure. Load / Unload: the two directions of the mechanical test. Displacement (mm): imposed displacement of the structure. Froce (N): measured applied force (the spreadsheet header contains this spelling; it means “Force”). Each direction has its own displacement and force columns.
Fig. 2H — Four-phase single-expansion oscillator output pressures.
Time (s): elapsed time for the pressure trace. A, B, C, D: measured pressures (kPa) at the four oscillator output ports, identified by the second header row under Pressure (kPa).
Fig. 2I — Single-expansion oscillator frequency versus input pressure.
Input pressure (kPa): supplied pneumatic pressure. Sealed outputs: oscillating frequency (Hz) with sealed oscillator outputs (no external actuator load). Driving actuators: oscillating frequency (Hz) while connected to pneumatic actuators. Both frequency columns fall under Oscillating Frequency (Hz).
Fig. 4D — Running gait and ground-contact states.
Time stamp (ms): video time in milliseconds. Cycle: position within the normalized gait cycle (dimensionless). L1, L2, L3, R1, R2, R3: binary contact/activity states of the left (L) and right (R) legs, numbered front to rear; 1 indicates active/contact and 0 indicates inactive/no contact. Aerial phase: binary indicator of the period when the robot is airborne (1) or not (0).
Fig. 4H — Duty factors of legs and aerial phase.
Seven measured gait cycles and their summary statistics. Sample: replicate number (1–7), followed by Mean and Standard deviation summary rows. L1, L2, L3, R1, R2, R3: fraction of a gait cycle in which the indicated leg is in ground contact (dimensionless, 0–1). Aerial phase: fraction of a gait cycle spent airborne (dimensionless, 0–1). N=7: number of cycles analyzed.
Fig. 4I — Comparison of robot speed and body length with published soft robots.
Reference number: numbered literature reference in the article, or This work for the present robot. Body length (cm): robot length in centimeters. Speed (cm/s): reported locomotion speed in centimeters per second.
Fig. 4J — Distance, speed, and acceleration during repeated runs.
Time (s): elapsed time since the start of a run; the time column appears separately for raw runs and processed statistics. Distance (m): distance traveled in each of five individual runs (five adjacent data columns under the same heading). Mean distance (m): average distance across the five runs. Distance standard deviation: standard deviation of distance (m). Mean speed (m/s): mean instantaneous speed across runs. Speed standard deviation: standard deviation of speed (m/s). Mean acceleration (m/s²): mean instantaneous acceleration. Acceleration standard deviation: standard deviation of acceleration (m/s²). N=5: five individual tests.
Fig. 4K — Speed at selected timestamps in a representative running test.
Time (s): elapsed time during the run. Speed (m/s): robot speed at the indicated time.
Fig. 5A — Running speed versus input pressure for three robot configurations.
Data are grouped under Double-expansion robot unloaded, Double-expansion robot loaded, and Single-expansion robot. Pressure (kPa): input pressure for that configuration. Sample 1, Sample 2, Sample 3: measured running speeds (m/s) in three replicate tests at each pressure. Mean: average of the three speeds (m/s). Standard deviation: standard deviation of those speeds (m/s). Loaded: double-expansion robot carrying the 257 g payload described in the paper; unloaded: without the payload. N=3: three tests per condition.
Fig. 5B — Eight output pressures of the double-expansion oscillator integrated into the running robot.
Time (s): time for each of the two separately recorded/aligned pressure groups. Port 1-A, 1-B, 1-C, 1-D: measured first-stage output pressures (kPa). Port 2-A, 2-B, 2-C, 2-D: measured second-stage output pressures (kPa). The number denotes the expansion stage; A–D denote the four sequential output phases.
Fig. 5C — Oscillator lifetime test.
Oscillating cycle (x1000): accumulated oscillation cycles, expressed in thousands of cycles. Oscillating frequency (Hz): measured oscillation frequency over the lifetime test.
Fig. S6 — Double-expansion oscillator pressures with external antagonistic actuators.
Time (s): time for the first- and second-stage pressure traces. Port 1-A, 1-B, 1-C, 1-D: first-stage output pressures (kPa). Port 2-A, 2-B, 2-C, 2-D: second-stage output pressures (kPa). A–D denote the four phases; the first-stage and second-stage ports are presented in separate groups. The connected first-stage and second-stage actuators are described in the Fig. S6 caption.
Fig. S9A — Force–displacement behavior of leg tips with and without springs.
Leg tip without spring / Leg tip with spring: two tested tip configurations. Displacement (mm): imposed leg-tip displacement. Force (N): measured contact force. Each configuration has its own displacement–force pair.
Fig. S10 — Published soft-robot body length and speed comparison.
Reference number: cited reference number or This work. Body length (cm): robot body length. Speed (cm/s): locomotion speed. The data support the supplementary comparison of locomotion speeds and robot sizes.
Fig. S11A — Pressure regulator setting versus actual robot input pressure.
Pressure setting at regulator (kPa): commanded upstream regulator pressure. Single-expansion robot / Double-expansion robot: corresponding measured Average input pressure (kPa) at the inlet of each robot, accounting for pressure loss in the supply tubing.
Fig. S11C — Air consumption versus input pressure.
Data are grouped for Double-expansion robot and Single-expansion robot. Average input pressure (kPa): measured mean pressure at the robot inlet. Flow rate (SLPM): air consumption in standard liters per minute. Each robot has a pressure–flow pair.
Fig. S11E — Oscillation frequency versus input pressure for running robots.
Double-expansion robot / Single-expansion robot: robot configurations. Average input pressure (kPa): mean inlet pressure. Oscillating frequency (Hz): corresponding oscillation frequency. Each configuration has its own pressure–frequency pair.
Fig. S12A — Output pressure traces of a double-expansion running robot.
Time (s): time axis for the two pressure groups. Port 1-A, 1-B, 1-C, 1-D: first-stage outlet pressures (kPa). Port 2-A, 2-B, 2-C, 2-D: second-stage outlet pressures (kPa). A–D distinguish the four output phases.
Fig. S12B — Output pressure traces of a single-expansion running robot.
Time (s): elapsed time. Port A, Port B, Port C, Port D: pressure (kPa) at each of the four oscillator outputs.
Fig. S12C — Double-expansion robot exhaust pressures versus input pressure.
Average input pressure (kPa): measured mean robot inlet pressure. Port A, Port B, Port C, Port D: exhaust pressure (kPa) associated with the indicated output phase/port in the double-expansion robot.
Fig. S12D — Single-expansion robot exhaust pressures versus input pressure.
Average input pressure (kPa): measured mean robot inlet pressure. Port A, Port B, Port C, Port D: exhaust pressure (kPa) associated with the indicated output phase/port in the single-expansion robot.
Fig. S13A — Pressure profiles used to estimate actuator work.
The first group contains Time (s) and four pressure traces (kPa) under the merged heading Pressure (kPa), representing oscillator output pressure profiles. The adjacent groups labeled Horizontal actuators and Vertical actuators each contain Time (s), Increasing side (kPa), and Decreasing side (kPa). Increasing side / Decreasing side: pressures in the inflating and deflating sides, respectively, of the antagonistic actuator pair during the selected motion. The first four pressure columns have no individual port labels in this worksheet.
Fig. S13C — Pressure–volume response of a horizontal leg actuator.
Inflating pressure (kPa): applied pressure during inflation. Relative volume (cm³): actuator volume change relative to its neutral state, in cubic centimeters.
Fig. S13D — Pressure–volume response of vertical leg actuators.
Inflating pressure (kPa): applied inflation pressure. Vertical top (cm³): volume change relative to neutral of the upper vertical actuator. Vertical bottom (cm³): volume change relative to neutral of the lower vertical actuator.
Fig. S14A — Power breakdown of the single-expansion robot.
Average input pressure (kPa): measured mean inlet pressure. Total input power (W): pneumatic input power. Exhaust power (W): pneumatic power remaining in exhaust flow. Internal power (W): power used internally by the oscillator and tubing. External power (W): estimated mechanical output power delivered to leg actuators.
Fig. S14B — Power breakdown of the double-expansion robot.
Average input pressure (kPa): mean inlet pressure. Total input power (W): pneumatic input power. Exhaust power (W): power lost in exhaust. External power (W): estimated mechanical power delivered to leg actuators. Internal power (W): internally consumed power associated with oscillator and tubing operation.
Fig. S14C — Pneumatic power delivery efficiency.
Two data groups, Single-expansion robot and Double-expansion robot, each contain Average input pressure (kPa) and Efficiency. Efficiency: dimensionless fraction of pneumatic input power converted to internal and external mechanical energy, i.e., the share not discharged as exhaust power (values 0–1, not percentages).
Fig. S14D — Mechanical output efficiency.
Two data groups, Single-expansion robot and Double-expansion robot, each contain Average input pressure (kPa) and Efficiency. Efficiency: dimensionless fraction of input pneumatic power delivered as external mechanical output power (values 0–1). Empty columns to the right do not represent additional variables.
Fig. S14E — Propulsive-phase mechanical output power.
Separate Single-expansion robot and Double-expansion robot groups. Average input pressure (kPa): mean inlet pressure. Power (W): estimated mechanical output power during the propulsive phase, when the legs swing backward against the ground; other phases are excluded.
Fig. S18 — Running speeds on different terrains and in water.
Terrain: testing environment (Indoor, Gravel, Soil, Lawn, Stairs, Mulch, Water Surface, or Water Bottom). Mean speed (m/s): average speed across three runs. Standard deviation (m/s): standard deviation of the three speeds. Sample 1 (m/s), Sample 2 (m/s), Sample 3 (m/s): speeds of the individual runs. N=3: three tests per environment. Water Surface: robot moving on the water surface; Water Bottom: underwater motion along the bottom.
Fig. S20 — Untethered robot performance using a CO₂ canister.
Three independently sampled time series: Time (s) with Input pressure (kPa) (pressure supplied to the robot); Time (s) with Oscillating frequency (Hz) (robot stepping/oscillation rate); and Time (s) with Speed (m/s) (instantaneous robot speed). The separate time columns correspond to different sampling rates and need not have matching row counts.
File: Data_file_S2.zip
Description: A ZIP archive containing four STL files used for monolithic 3D printing of standalone pneumatic oscillators and complete soft running robots.
Files:
- Double-expansion oscillator.stl: A standalone pneumatic oscillator using a double-expansion mechanism.
- Double-expansion robot.stl: A complete soft running robot with an integrated double-expansion pneumatic oscillator.
- Single-expansion oscillator.stl: A standalone pneumatic oscillator using a single-expansion mechanism.
- Single-expansion robot.stl: A complete soft running robot with an integrated single-expansion pneumatic oscillator.
These files provide the 3D geometries needed to fabricate the corresponding oscillators and robots.
Code/software
Data file S1 is an Excel workbook (.xlsx) that can be opened using Microsoft Excel or the free, open-source software LibreOffice Calc. Data file S2 is a ZIP archive containing four 3D design files in STL format. The STL files can be viewed using free software such as FreeCAD or Blender. No additional packages or custom scripts are required to access or view the files.
Access information
Other publicly accessible locations of the data:
- None.
Data was derived from the following sources:
- None. All experimental data and 3D design files were generated by the authors.
The pneumatic oscillators and soft running robots were designed using CAD software and fabricated monolithically using a modified desktop fused filament fabrication (FFF) 3D printer with thermoplastic elastomer (TPE) filament. Continuous Eulerian printing paths were used to fabricate airtight structures with embedded pneumatic control circuits.
Experimental data were collected through pressure and airflow measurements, force testing, high-speed video recordings, and locomotion experiments. Pressure sensors and a mass flow meter were used to characterize pneumatic performance, while a force tester was used to measure mechanical forces. High-speed cameras and video analysis were used to characterize oscillation, gait, and locomotion. Running speeds were measured using video tracking or laser sensors, depending on the experiment. Tests were conducted under different supply pressures and across various terrestrial and aquatic environments.
The dataset contains tabulated experimental results and 3D CAD design files used in the study. Further details of the experimental methods are provided in the associated research article and its Supplementary Materials.
