A flexible sensor technology for the distributed measurement of interaction pressure

Abstract

We present a sensor technology for the measure of the physical human-robot interaction pressure developed in the last years at Scuola Superiore Sant'Anna. The system is composed of flexible matrices of opto-electronic sensors covered by a soft silicone cover. This sensory system is completely modular and scalable, allowing one to cover areas of any sizes and shapes, and to measure different pressure ranges. In this work we present the main application areas for this technology. A first generation of the system was used to monitor human-robot interaction in upper- (NEUROExos; Scuola Superiore Sant'Anna) and lower-limb (LOPES; University of Twente) exoskeletons for rehabilitation. A second generation, with increased resolution and wireless connection, was used to develop a pressure-sensitive foot insole and an improved human-robot interaction measurement systems. The experimental characterization of the latter system along with its validation on three healthy subjects is presented here for the first time. A perspective on future uses and development of the technology is finally drafted.

Figure 1.

Overview of the PSP1: (a) specific scheme of the 1 × 8 array of sensitive elements of the PSP 1.0 and PSP1.1; (b) scheme of the transduction principle; (c) 3D design of the PSP1.1 (adapted from [37]).

Figure 2.

Cross section of the PSP1.1 (adapted from [37]).

Figure 3.

Finite element simulation of PSP1.1: (a) un-deformed structure; the rigid flat indenter is transparent brown, the silicone structure is grey and the PCB is green; (b) total deformation representation; (c) total stress representation (adapted from [37]).

Figure 4.

Force (or pressure) vs. output voltage of the PSP1 1 × 8 array of sensitive elements: (a) PSP1.0; (b) PSP1.1 (adapted from [33,37]. (a) is a slightly adapted reprinted graphics from [33], ©2011, with permission from Elsevier).

Figure 4.

Force (or pressure) vs. output voltage of the PSP1 1 × 8 array of sensitive elements: (a) PSP1.0; (b) PSP1.1 (adapted from [33,37]. (a) is a slightly adapted reprinted graphics from [33], ©2011, with permission from Elsevier).

Figure 5.

Overview of the second-generation PSP sensitive element: (a) dimension of the sensitive element, (b) transduction principle (adapted from [39], ©2011 IEEE. Reprinted with permission).

Figure 6.

Cross section of the PSP2 silicone cover; the geometrical parameters are: cover thickness T, height of the curtain H1, pyramidal frustum height H2, square base size B1, square top-face size B2.

Figure 7.

3D FE simulations of PSP2.1: (a) simulation environment: in blue the rigid indenter, in grey the silicone structure, in green the PCB; (b) map of the total deformation, (c) cross-section of the pyramidal frustum showing the sinking effect.

Figure 8.

Characterization of the sensitive element of PSP2.0: (a) quasi-static force-to-deformation characterization; (b) quasi-static force-to-voltage curve (adapted from [39], ©2011 IEEE. Reprinted, with permission).

Figure 9.

Results of the PSP2.1 characterization: (a) quasi-static force-to-deformation loading-unloading, averaged over three iterations (black line is the loading phase, grey line is the unloading phase); (b) all fitting curves of loading-unloading cycles at seven different levels of loading speed (namely, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 1 mm/s); (c) quasi-static force-to-voltage curve (blue dots are experimental data, black line is the smoothing spline).

Figure 10.

The NEUROExos platform equipped with two PSPs 1.0. (a) PSP placement onto inner-side of exoskeleton inner-shells; front (b) and lateral (c) view of a subject wearing the NEUROExos (adapted from [33], ©2011, with permission from Elsevier).

Figure 11.

NEUROExos joint trajectory, front- and back-side PSP1.0 pressure profiles during a prototypical rehabilitation task, with and without the user reaction. Top panel reports: the reference trajectory of the rehabilitation task (dashed line), the “no action” trajectory (gray line), and the “pre-defined action” performed by the subject (black line). Middle and bottom panels report respectively front- and back-side PSP1.0 pressure profiles: the “no action” condition is the gray line, the “pre-defined action” is the black line. Pressure profiles are averaged over ten sinusoidal motions and reported along with standard-deviation contour (dotted line). We assume that the elbow is fully extended when the joint angle is equal to zero (figure adapted from [33], ©2011, with permission from Elsevier).

Figure 12.

(a) Overview of the LOPES exoskeleton; (b) right-leg thigh cuff sensorized with six PSPs 1.1; (c) schematic view of the right-leg sensorized cuff (adapted from [37]).

Figure 13.

Test of PSP1.1 on the LOPES: profiles of right hip flexion-extension angle, total interaction force measured by the load cell, and the force estimated by three of the six PSPs (i.e., “Front 1”, “Front 2”, “Rear 1”). Data are shown for two conditions: transparent and viscous field (figure adapted from [37]).

Figure 14.

PSP2.0-based pressure-sensitive insoles. (a) Overview of the pressure-sensitive insole on the bench; (b) two pressure-sensitive insoles integrated into normal sneaker shoes.

Figure 15.

Gait phases recognition through the pressure-sensitive insole for both left (top panel) and right (bottom panel) feet.

Figure 16.

New PSP2.1-based LOPES sensorized cuffs. (a) Overview of the 8 × 4 and 4 × 4 sensitive arrays; (b) sensorized thigh cuff; (c) sensorized shank and ankle cuffs; (d) overview of the LOPES with all of the six cuffs endowed with PSPs 2.1.

Figure 17.

Hip kinematic and dynamic variables for Subject #1: right-leg hip flexion-extension joint angle and torque, and total force recorded by front- (F-S) and back-side (B-S) thigh-cuff PSPs. Data are averaged over 20 gait cycles (solid line), and shown along with the standard deviation contour (shadowed), for three conditions: (a) gait velocity is 2.5 km/h, with assistive torque; (b) gait velocity is 4 km/h, with assistive torque; (c) gait velocity is 4 km/h without assistive torque.