Development of the Ultralight Hybrid Pneumatic Artificial Muscle: Modelling and optimization

Abstract

Pneumatic artificial muscles (PAMs) are one of the key technologies in soft robotics, and they enable actuation in mobile robots, in wearable devices and exoskeletons for assistive and rehabilitative purposes. While they recently showed relevant improvements, they still present quite low payload, limited bandwidth, and lack of repeatability, controllability and robustness. Vacuum-based actuation has been recently demonstrated as a very promising solution, and many challenges are still open, like generating at the same time a large contraction ratio, and a high blocking force with enhanced axial stiffness. In this paper, a novel Ultralight Hybrid PAM (UH-PAM), based on bellow-type elastomeric skin and vacuum actuation, is presented. In particular, open-cell foam is exploited as a structural backbone, together with plastic rings, all embedded in a thin skin. The design and optimization combine numerical, analytical, and experimental data. Both static and dynamic analysis are performed. The weight of the optimized actuator is only 20 g. Nevertheless, a contraction ratio up to 50% and a maximum payload of 3 kg can be achieved. From a dynamic point of view, a rise time of 0.5 s for the contraction phase is observed. Although hysteresis is significant when using the whole contraction span, it can be reduced (down to 11.5%) by tuning both the vacuum range and the operating frequency for cyclic movements. Finally, to demonstrate the potentiality of this soft actuation approach, a 3 DoFs Stewart platform is built. The feasibility of performing smooth movements by exploiting open-loop control is shown through simple and more complex handwriting figures projected on the XY plane.

Fig 1

Concept design of the proposed pneumatic artificial muscle: (a) bellow-type skin; (b) rigid rings; (c) vacuum line; (d) open-cell foam cylinders.

Fig 2

Design parameters: (a) a bellow structure at the initial state with null applied vacuum; (b) top view of bellow structure; (c) a single segment in the contracted state induced by vacuum.

Fig 3

FEM results: a) simulation setup of half structure, where planar symmetry was exploited; (b) deformation of a simple cylindrical skin structure with -0.62 kPa of applied vacuum; c) deformation of bellow Type 3 with -0.21 kPa of applied vacuum; (d) pressure versus deformation graphs for all bellow types; (e) deformation versus strain energy graphs for all bellow types. In (b) and (c): the color legends show deformation values in mm; the gray models show the relative meshes used for the simulations.

Fig 4

Fabrication procedure: For the two halves of the skin (a) pouring of specific elastomer in 3D printed mold (1); (b) mold cover (2) assembling; (c) after curing, the skin is detached from the molds; (d) foam elements are bonded using a silicone epoxy; (e) foam segments and Plexiglas^© rings are integrated into one-half skin by a mold; (f) mold (3) is used to host the second half skin and for bonding with the assembled structure.

Fig 5

Cyclic compression test results: (a) images showing the compression testing at different strain phases; (b) the experimental results of loading/unloading cycles for each case; (c) linearly interpolated lines (LSM) for each case.

Fig 6

Experimental results of static and quasi-static characteristics of the UH-PAM: (a) contraction ratio versus pressure characteristics; (b) comparison between the experimental (pink) and theoretical (blue) blocking force; (c) sequence of images showing lifting of 1kg lifting by negative pressure; (d) comparison of contraction and elongation ratio according to the attached weight from 1 to 3 kg; (e) contraction ratio versus pressure graph according to the attached weight from 1 to 3 kg.

Fig 7

Dynamic characterization results: (a) step response from 0 to 6 V; (b) stepwise response with random time spans for the pressure versus the contraction ratio; (c) block diagram for open-loop control of the UH-PAM.

Fig 8. Configuration of the 3DoFs Stewart platform integrating n.3 UH-PAMs, a fixed joint, a rigid cone, and the AURORA^® tracking system with the 3D coordinate system.

Fig 9. Scheme of the 3D coordinate of the projected vector r for the orientation of the Stewart platform.

Fig 10

Results of 3DoFs Stewart platform performance: (a) workspace by joy stick in 3D; (c) workspace in yz plane; (c) workspace in xy plane; (d) tracked travel of circle drawing in 3D; (e) tracked travel of triangle drawing in 3D; (f) tracked travel of rectangle drawing in 3D; (g) tracked travel of circle drawing in 2D; (h) tracked travel of triangle drawing in 2D; (i) tracked travel of rectangle drawing in 2D.