A soft, self-sensing tensile valve for perceptive soft robots

Abstract

Soft inflatable robots are a promising paradigm for applications that benefit from their inherent safety and adaptability. However, for perception, complex connections of rigid electronics both in hardware and software remain the mainstay. Although recent efforts have created soft analogs of individual rigid components, the integration of sensing and control systems is challenging to achieve without compromising the complete softness, form factor, or capabilities. Here, we report a soft self-sensing tensile valve that integrates the functional capabilities of sensors and control valves to directly transform applied tensile strain into distinctive steady-state output pressure states using only a single, constant pressure source. By harnessing a unique mechanism, "helical pinching", we derive physical sharing of both sensing and control valve structures, achieving all-in-one integration in a compact form factor. We demonstrate programmability and applicability of our platform, illustrating a pathway towards fully soft, electronics-free, untethered, and autonomous robotic systems.

Fig. 1. Soft self-sensing tensile valve (STV) transducing strain into manageable proportional output pressures.

a Photograph of the fabricated STV. Scale bar = 3 cm. b Schematic diagrams of a projected STV and finite element analysis (FEA) cross-sectional results when tensile strain is not applied ε = 0 (i, ii) and is fully applied ε = ε_max (iii, iv). The valve consists of an elastomeric inner tube that is tied with wrapping helical yarns (WHYs), an enclosing elastomeric outer tube, and connectors at the ends that regulate the path of the airflow. The air from a pressure supply flows inside the STV in the following directional path: inlet (P_s) - inlet channel - chamber (P_ch) - outlet channel - outlet (P_atm). c An analogous electrical circuit that represents the STV pressure control mechanism. d Experimental and computational fluid dynamics (CFD) simulation results of the output chamber pressure to supply pressure ratio P_ch/P_s as a function of the normalized strain ε/ε_max. Inset: schematic of a soft pneumatic actuator at P_ch/P_s = 0 (i), P_ch/P_s = 0.5 (ii), and P_ch/P_s = 1 (iii). e Operational photographs of the STV-controlled soft actuator when the tensile strain ε = 0 (i) and ε = ε_max (ii). Insets: close-up images of the STV. Scale bars = 5 cm.

Fig. 2. Programming output chamber pressure curves.

a Schematic of the design parameters (i) and cross-sectional description (ii). b FEA cross-sectional results with contours of the normalized elastic strain energy density W/W_max upon tensile strain and microcomputed tomography (micro-CT) images at ε = ε_max for n = 1 (i), n = 2 (ii), n = 3 (iii), and n = 4 (iv). Scale bars = 1 mm. c Normalized chamber pressure P_ch/P_s plotted against normalized strain ε/ε_max for STVs with different numbers of WHYs n. d Maximum strain ε_max as a function of the cyclic pitch of WHY p (n = 2, L₀ = 80 mm). e Normalized chamber pressure curves as a function of extension length ΔL with different initial lengths traced by the center of WHY L₀ (n = 2, p = 10 mm). f Chamber pressure P_ch plotted against normalized strain ε/ε_max for different supply pressures P_s (n = 2, p = 10 mm, L₀ = 80 mm). g Schematics of the STVs with original (top) and inversed (bottom) connections, and their normalized chamber pressure P_ch/P_s are plotted against the normalized strain ε/ε_max in (h) (n = 2, p = 10 mm, L₀ = 80 mm).

Fig. 3. Untethered and electronics-free soft gripper.

a A photograph of the soft gripper picking up a pinecone outdoors. b Details of the soft gripper. c Photographs of the soft gripper at normalized strain ε/ε_max = 0.5 (left) and ε/ε_max = 1 (right). The curvature of the soft actuators is continuously controlled by extending the STV (n = 2, p = 10 mm, L₀ = 80 mm) with the handle. d Holding force of the gripper plotted against the normalized strain ε/ε_max at supply pressure P_s = 60 kPa. e Maximum holding force of the gripper plotted against the supply pressure P_s. f Gripping demonstrations at P_s = 60 kPa: a cabbage (i), a tree bark (ii), a chestnut bur (iii), a tennis ball (iv), a pinecone (v), and a potato chip (vi). Scale bars = 5 cm. g Photographs of the soft gripper handling a pinecone near a welding arc (which can affect electronic circuits by electromagnetic interference), and (h) coral reef under water.

Fig. 4. Autonomous and self-adaptive soft exosuit.

a Overview of the soft exosuit. A photograph of the soft exosuit on a user (i) and details of the soft exosuit (ii). The STV (n = 2, p = 10 mm, L₀ = 80 mm) is connected in inverse mode, where the produced normalized chamber pressure P_ch/P_s is a decreasing function of the normalized strain ε/ε_max (iii). b Schematics of the exosuit and the STV at the start and end elbow angles. The STV reaches the maximum strain ε = ε_max at θ = 0° (i) and releases strain ε = 0 to its original length at θ = 60° (ii). c Normalized chamber pressure P_ch/P_s plotted against the elbow angle θ. Insets: schematic describing the elbow angle when θ = 0° (i), θ = 30° (ii), and θ = 60° (iii). d Sequential photographs of a mannequin with an applied elbow torque of 1 Nm by a motor and subsequent gripping of a dumbbell (0.5 kg). e Results with the mannequin with our soft exosuit from the same procedures in d. f Assistive torque T_z plotted against elbow angle θ. The constant P_ch (blue line) represents data with our soft exosuit, but without STV control (constant P_ch = 80 kPa).