3D printing process
Commercially available flexible filament eFlex (TPU, shore hardness 87A, eSUN) was used as the 3D printing material, and all experiments were conducted with robots made of this material. Other flexible materials including eTPU-95A (TPU, shore hardness 95A, eSUN), NinjaFlex (TPU, shore hardness 85A, NinjaTek), and eLastic (TPE, shore hardness 85A, eSUN) were also tested and it was confirmed that all of the tested materials can be used. Open-source slicing software Ultimaker Cura (Ultimaker BV) was used to prepare stereolithography (STL) files and they were imported into DIY 3D printer CORE 200 (Making Tool) using a 0.4 mm nozzle. For stable printing, the print speed was set as slow as 20 mm/s, and it took about 4 hours to print the six-module dual-origami and about 9 hours to print the dual-origami two-finger gripping unit. For stiffer material eTPU-95A, 40 mm/s of print speed could be applied, and the printing time was halved. Printer performance affected the consistency of soft robots’ motion because a small error in strain-limiting layer thickness (related to bending stiffness~\(t^3\)) significantly changed the dominance of deployment and bending.
 
Characterization and Measurement:
For the parametric study, we prepared five six-module dual-origami soft fluidic bending actuators for each geometric parameter. Red or blue dots were marked with a pen on the side edge of the modules. A protruded diamond shaped crease line loop at the bottom was surrounded and held by 3d printed rigid parts, and then these rigid parts were fixed to an aluminum optical breadboard. Dual-origami soft fluidic bending actuators were installed vertically to the gravity direction to ensure that gravity does not affect the deployment. For every specimen, photographs were taken for applied pressure from 0 to 200 kPa with 10 kPa increments, which was precisely controlled by a pressure regulator (RVUM, PISCO). The photographs were analyzed with MATLAB (MathWorks). The center positions of the marked circles were automatically found and characterized based on the color information. The angle between adjacent modules (\(\theta_{ori}\) and \(\theta_{free}\)) was derived by calculating the angle between straight lines connecting points, and effective layer length (\(L\)) was derived using cubic Hermite spline of points at strain-limiting layer side. The bending angle (\(\phi\)) was derived by calculating the angle between the axis in the height direction and the line connecting the floor and the end-tip trajectory. For the gripping force measurement, a regulated constant pressure (from 100 kPa to 220 kPa with 20 kPa increment for each experiments) was applied to the two-finger gripping unit holding a cylinder located at the end of the rail connected to a load cell (333FDX, KTOYO) and a linear actuator (P16-P, Actuonix Motion Devices, Inc.). To estimate the scale of friction, the rail without external gripping force was also pulled and the friction force was measured to be as low as below 50 gf.
 
Finite Element Analysis 
Finite element analysis was conducted using FEA software ABAQUS (Dassault systems). Uniaxial tensile test for eFlex was performed following iso standard ASTM D412, and the result was imported into ABAQUS as a material property. All simulation conditions were set identical to the experimental conditions; the protruded diamond shaped crease line was set as a fixed boundary condition, and the contacts between facets were considered using the ‘general contact condition’. Nodes corresponding to the points marked in the actual experiment were selected and their unique nodal displacements were collected to calculate the angle between adjacent modules, deployment ratio, and bending angle.
 
Gripping tasks
Figure 14 shows the soft gripper successfully conducting gripping tasks designated at the manipulation competition of Robosoft 2021. The first task requires strength to grip a heavy and large fruit (1-2 kg), and the second task requires delicacy to grip a small sized delicate fruit. Korean radish weighing 1.3 kg and raspberry were chosen respectively. The third and fourth tasks were to pick a coin an insert it into a coin slot, requiring precision and dexterity respectively. The Korean 100-won coin (24 mm diameter) was chosen and the coin slot was fabricated as 4 mm X 27 mm using an acrylic plate.