Pump Up the Jam: Granular Media as a Quasi-Hydraulic Fluid for Independent Control Over Isometric and Isotonic Actuation

Abstract

Elastomer-granule composites have been used to switch between soft and stiff states by applying negative pressure differentials that cause the membrane to squeeze the internal grains, inducing dilation and jamming. Applications of this phenomenon have ranged from universal gripping to adaptive mobility. Previously, the combination of this jamming phenomenon with the ability to transport grains across multiple soft actuators for shape morphing has not yet been demonstrated. In this paper, the authors demonstrate the use of hollow glass spheres as granular media that functions as a jammable "quasi-hydraulic" fluid in a fluidic elastomeric actuator that better mimics a key featur of animal musculature: independent control over i) isotonic actuation for motion; and ii) isometric actuation for stiffening without shape change. To best implement the quasi-hydraulic fluid, the authors design and build a fluidic device. Leveraging this combination of physical properties creates a new option for fluidic actuation that allows higher specific stiffness actuators using lower volumetric flow rates in addition to independent control over shape and stiffness. These features are showcased in a robotic catcher's mitt by stiffening the fluid in the glove's open configuration for catching, unjamming the media, then pumping additional fluid to the mitt to inflate and grasp.

Keywords: fluidic elastomer actuators; granular flow; granular jamming; soft robotics; variable stiffness.

Figure 1

Flow behavior, selection, and demonstration of granular materials for quasi‐hydraulic fluidic elastomeric actuator: A) flow‐curves of solid glass spheres (SGSs), coffee grounds (CG), and hollow glass spheres (HGSs) measured after three pre‐shear events and fitted with the Ostwald de Waele power law model (sample size n = 3). B) Plot of the plunger displacement as a function of time for a pump extruding HGSs, SGSs, and CGs from 10 mL syringes through nozzle diameters of d_i = 10 mm and d_i = 6.4 mm (sample size n = 3). C) SEM images of HGSs taken after a syringe pump extruded them back and forth through a d_i = 6.4 mm tapered nozzle and five pumping cycles. D) A 3D render of a robotic catching mitt made of quasi‐hydraulic fluidic elastomeric actuators (QH‐FEAs) with HGSs.

Figure 2

Design and operation of fluidic device. A) Flow schematic of fluidic device in configuration to inflate a QH‐FEA. B) Flow schematic of fluidic device in configuration to jam a QH‐FEA. Note that Solenoid Valve 2 is opened between states A and B to release the buildup of air pressure within the QH‐FEA. C) Photographs of cylindrical QH‐FEAs in various inflation states: (from left to right) ΔP < 0 evacuated, ΔP = 0 filled to equilibrium, ΔP > 0 filled beyond equilibrium. D) Photographs of filled and then jammed cylindrical QH‐FEAs demonstrating ability to bear 1 kg load. (From left to right) ΔP = unjammed HGSs, unable to support weight, ΔP < 0 induced jammed state, ΔP < 0 jammed state supporting weight.

Figure 3

Rigidity modulation of QH‐FEA elastomer‐granular fluid composites. A) 3D render of a cylindrical QH‐FEA for compression testing. B) Stress–strain curves of cylindrical QH‐FEAs when empty (ΔP = 0 kPa), filled with hollow glass spheres (ΔP = 0, −14, and −70 kPa) and filled with coffee grounds (ΔP = −70 kPa) tested in compression (sample size n = 8). C) A 3D render of a cylindrical FEA for shear testing with adapters that couple to the rheometer. D) Torque sweeps of cylindrical QH‐FEAs when empty (ΔP = 0) and when filled with hollow glass spheres (ΔP = 0, −14, and −70 kPa) (sample size n = 3). E) 3D render of a bending QH‐FEA. F) Bending angle absolute deflection of bending QH‐FEA in different inflation configurations under 50 g load while empty (ΔP = 0 kPa for the relaxed state) and while filled with HGSs (ΔP = 0 kPa and −70 kPa) (sample size n = 3). Negative bending angles represent working upward against gravity, positive bending angles represent downward bending.

Figure 4

Design and control of a soft robotic catching mitt. A) Series of time lapse photographs of a tennis ball being dropped from a height of 30 cm onto an empty robotic catching mitt inflated with ΔP = 2.5 ± 0.2 kPa and finger deflection 50 ± 0.5 mm. B) Series of time lapse photographs of a tennis ball being dropped from a height of 30 cm onto a robotic catching mitt filled with HGSs (ΔP = −70 kPa) and inflated with and finger deflection 50 ± 0.5 mm. C) After time = 5 s, the pumping direction was reversed to inflate the QH‐FEAs and grasp the tennis ball. Scale bar of 10 cm is applicable to all subfigures.