WO2019241671A1

WO2019241671A1 - Modular artificial skeletal muscle actuators and exoskeletons powered thereby

Info

Publication number: WO2019241671A1
Application number: PCT/US2019/037250
Authority: WO
Inventors: Ahad BEHBOODI; Samuel C.K. LEE; Stuart A. BINDER-MACLEOD; Henry Wright
Original assignee: Individual
Current assignee: Individual
Priority date: 2018-06-14
Filing date: 2019-06-14
Publication date: 2019-12-19
Anticipated expiration: 2020-12-14

Abstract

A modular artificial skeletal muscle (MASM) actuator having a plurality of dielectric elastomer units assembled in series or in parallel. A proximal anchoring member is connected to a proximal end of at least one proximal dielectric elastomer unit and is configured for attachment to a proximal anchor point relative to a user. A tension member has a proximal end attached to a distal end of at least one distal dielectric elastomer unit and a distal end configured for attachment to a distal anchor point relative to the user. A controller is configured to cause simultaneous contraction or expansion of plurality of dielectric elastomer units, which contraction or expansion causes displacement of the tension member. A plurality of MASM actuators may be attached to an exoskeleton attached about an ankle of a user and controlled for therapeutic purposes, such as for assisting pediatric patients with cerebral palsy.

Description

MODULAR ARTIFICIAL SKELETAL MUSCLE ACTUATORS AND

EXOSKELETONS POWERED THEREBY

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority from U.S. Provisional Application Ser. No. 62/684,908, filed June 14, 2018, and from U.S. Provisional Application Ser. No.

62/745,512, filed October 15, 2018, both titled "AMPEX, Artificial Muscle Powered Exoskeleton," the contents of both of which are hereby incorporated by reference.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

This invention was made with government support under Grant Number U54-GM 104941 awarded by the National Institute of Health (NIH) to Delaware-CTR ACCEL (Accelerating Clinical and Translational Research). The government has certain rights in the invention.

BACKGROUND OF THE INVENTION

To improve the acceptability and performance of rehabilitation robots, development of actuator technology with natural feel is highly desirable. For exoskeleton applications, it is desirable for the actuator to have force and displacement generation comparable to that of mammalian skeletal muscle. Actuators that are soft, noiseless, have linear actuation, and produce comparable force and displacement to skeletal muscle are referred to as artificial skeletal muscles.

Soft electroactive polymer actuators, such as dielectric elastomer actuators (DEAs), are fast, efficient and offer a lightweight and acoustically noiseless alternative to DC motor actuators used in conventional rehabilitation robotic systems. Dielectric elastomer actuators comprise a compliant capacitor, namely an elastic dielectric sandwiched between two mechanically compliant electrodes

There are two main DEA configurations: 1) freestanding DEAs and 2) DEAs with rigid frames. Rigid frames may not be ideal where soft actuators are preferable. Consequently, in artificial skeletal muscle applications, freestanding configurations offer more natural feel to the design. Various freestanding configurations are known, including stacked, folded, helical, membrane, and spring roll DEAs, as shown in FIG. 1. Stacked DEAs have shown the most similarity to human skeletal muscle compared with other linear DEA configurations, and have shown a capability of producing 32N of isometric force and 15% strain while lifting a l. lKg tensile load.

A commercially available Stacked DEA, depicted in FIG. 2, is manufactured by CTsystems of Dubendorf, Switzerland. SUMMARY OF THE INVENTION

One aspect of the invention comprises a modular artificial skeletal muscle (MASM) actuator. The MASM actuator includes a plurality of dielectric elastomer units assembled in series or in parallel. A proximal anchoring member is connected to a proximal end of at least one proximal dielectric elastomer unit and configured for attachment to a proximal anchor point relative to a user. A proximal end of a tension member, such as a flexible, non-extendable cable, is attached to a distal end of at least one distal dielectric elastomer unit. A distal end of the tension member is configured for attachment to a distal anchor point relative to the user. A controller attached to a power source and to the plurality of dielectric elastomer units is configured to cause simultaneous contraction or expansion of plurality of dielectric elastomer units, which contraction or expansion causes displacement of the tension member. The MASM actuator may have a flexible, low friction covering, which also acts as an electrical insulator, disposed over an external surface of the plurality of dielectric elastomer units. A self-sensing circuit connected to the controller may be configured to detect a relative longitudinal state of the plurality of at least one sensed dielectric elastomer unit in the MASM and to provide feedback to the controller. The controller, which may be a PID controller, may be configured to control contraction or expansion of the plurality of dielectric elastomer units based upon the detected relative longitudinal state of the at least one sensed dielectric elastomer unit. In other embodiments, an electromyographic (EMG) signal processing interface may be connected to the controller and to one or more EMG sensors configured to detect electrical activity of a living muscle, and the controller may be configured to control contraction or expansion of the plurality of dielectric elastomer units based upon a signal from the EMG signal processing interface.

Another aspect of the invention comprises a powered exoskeleton configured to be worn by a user, comprising one or more modular MASM actuators as described above. The powered exoskeleton comprises a frame including at least an proximal portion configured to be removably attached to the user proximal a joint of the user, and a distal portion configured to be removably attached to the user distal of the joint of the user. The proximal and distal portions configured to move relative to one another with at least one degree of freedom. Each MASM actuator proximal anchoring member is attached to the proximal portion of the frame at the proximal anchor point. Each MASM actuator tension member distal end is attached to the distal portion of the frame at the distal anchor point. Each tension member has an adjustable length between the proximal anchor point and the distal anchor point. The

displacement of the respective tension member caused by contraction or expansion of each MASM actuator causes corresponding displacement of the distal portion of the frame relative to the proximal portion of the frame.

One or more tension members in the exoskeleton system may be connected to a displacement magnification system that converts a first amount of displacement of the tension member into a greater amount of displacement of the distal portion of the frame relative to the proximal portion. In certain embodiments, the frame may be configured for removable attachment to a pediatric user.

In one embodiment, the frame is configured for placement about an elbow joint of the user, such as having a proximal portion configured to be anchored about a shoulder of the user and a distal portion comprising a cuff configured to be anchored to a forearm of the user. In another embodiment, the frame may be configured for placement about an ankle joint of the user.

The controller may comprise a control system including one or more feedback sensors and is configured to activate preset sequences of the plurality of MASM actuators in response to specific detected system state conditions. In an embodiment in which the frame is configured for placement about an ankle joint of the user, the system may include at least one plantarflexor MASM actuator configured to augment ankle plantarflexor muscle function and at least one dorsiflexor MASM actuator configured to augment ankle dorsiflexor muscle function. The system may further comprise at least one inversion MASM actuator configured to augment ankle inversion muscle function and at least one eversion MASM actuator configured to augment ankle eversion muscle function. The controller may include a control system including a sensor configured to detect different phases of ankle motion during a walking gait of the user and to activate preset sequences of the plantarflexor, dorsiflexor, inversion, and eversion MASM actuators in synchrony with the detected phases of the walking gait. The sensor configured to detect different phases of ankle motion during the walking gait may comprise a gyroscope configured to detect bilateral shank angular velocity of the user for determining gait phase transitions and a plurality of stimulators for activating the MASM actuators.

In one embodiment, the powered exoskeleton is configured to

rehabilitate a gait of a pediatric user with cerebral palsy, and the controller is configured with an algorithm for detecting gait phase based upon z axis bilateral shank angular velocity or a sum of bilateral shank angular velocity about three axes and which includes detection criteria tailored to gait phase transitions of children with cerebral palsy. The detection criteria tailored to gait phase transitions of children with cerebral palsy may include a delay in Initial Swing (ISw) detection until after a predetermined percentage of elapsed gait cycle, a delay in Mid-Swing (MSw) detection until a predetermined number of samples after ISw, a peak detection threshold for ISw set to a predetermined percentage of a smallest detected peak in a predetermined number of previous cycles, and a peak detection threshold for Terminal Swing (TSw) set to a predetermined percentage of a smallest detected valley in a predetermined number of previous cycles, wherein the peak for ISw may be evaluated as a sum of the bilateral shank angular velocity about the three axes, if desired for better resolution of the peak.

In one ankle embodiment, the distal portion of the frame may comprise a footplate configured for positioning beneath a foot of the user. The proximal portion of the frame may be configured for positioning about a lower leg of the user and further comprises an exoskeletal connector connecting the footplate to a connection point on the proximal portion of the frame, and configured to extend along the lower leg of the user. The plurality of dielectric elastomer units of each MASM actuator are disposed in a pocket that is attached to a textile member and configured to be removably anchored to the proximal portion of the frame. Each respective tension member may be confined for some portion of its length within a relatively low friction conduit in the textile member. The proximal portion of the frame may comprise a first portion configured to be mounted proximal of a knee of the user, a second portion configured to be mounted distal of the knee of the user, the first and second portions connected to one another on opposite sides of the knee of the user with opposite exoskeletal knee joints. The tension member connected to a dorsiflexor MASM actuator configured to augment ankle dorsiflexor muscle function may connects to a housing mounted to an adjustable, cable-reinforced instep strap connected to opposite sides of the footplate.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates various types of known DEA structures.

FIG. 2 illustrates an exemplary CT-DEA unit.

FIG. 3A illustrates an exemplary 1x4 modular artificial skeletal muscle (MASM) actuator.

FIG. 3B illustrates an exemplary 3x4 modular artificial skeletal muscle (MASM) actuator.

FIG. 3C depicts an isolated view of the exemplary 1x4 MASM actuator of

FIG. 3A.

FIG. 3D illustrates the exemplary 1x4 MASM actuator of FIG. 3C, encased in a flexible, electrically insulating, minimal friction covering.

FIG. 4 illustrates a plurality of exemplary clamping units.

FIG. 5 illustrates an exemplary clamping unit housing.

FIG. 6A illustrates an exemplary electrical circuit comprising a DEA. FIG. 6B illustrates an exemplary simplified circuit used for modeling DEA behavior.

FIG. 7 illustrates the parts of an exemplary set of curves relevant to self- sensing applications.

FIG. 8 illustrates an exemplary sensing + actuating signal curve relevant to self-sensing applications.

FIG. 9 illustrates exemplary curves relevant to self-sensing applications.

FIG. 10 schematically illustrates measured and fitted curves relevant to a self-sensing application.

FIG. 11 schematically illustrates functional and electrical representations for an exemplary DEA actuator.

FIG. 12A schematically illustrates an exemplary elbow exoskeleton embodiment.

FIG. 12B schematically illustrates an exemplary displacement magnifier embodiment.

FIGS. 13A-13D schematically illustrate a first exemplary ankle exoskeleton embodiment.

FIGS. 14A and 14B schematically illustrate an exemplary footplate.

FIGS. 15A-15D schematically illustrate a second exemplary ankle exoskeleton embodiment.

FIG. 16A illustrates a typical gait cycle for a healthy individual as measured by an IMU mounted on the shank of a subject.

FIG. 16B illustrates typical gait cycles for a typically developing (TD) child, a child with cerebral palsy (CP), and a curve showing the sum of all three axes of a gyroscope IMU mounted on a shank of a subject.

FIG. 17 schematically illustrates a control system for use with a gait phase detection (GPD) equipped controller for an MASM powered exoskeleton.

DETAILED DESCRIPTION OF THE INVENTION

Aspects of the invention comprise the use of modular artificial skeletal muscle (MASM) actuators as described herein, each of which comprises a plurality of dielectric elastomer actuator (DEA) units assembled in series or in parallel. While the examples as presented herein use stacked DEA, specifically stacked DEA made by CTsystems (CT25.0-15-15-71, Compliant Transducer Systems, Dubendorf,

Switzerland), depicted in FIG. 2, referred to specifically herein as CT-SDEA, the invention is not limited to any particular configuration of DEA (e.g. stacked, folded, etc.), nor is it limited to any particular manufacturer. Thus, even when referred to as CT-SDEA in various places herein, it should be understood that the other types of DEA muscle may also be applicable to any of the embodiments as discussed herein. Unless expressly indicated, use of DEA or SDEA or CT-SDEA should herein should not be interpreted as a limitation.

As shown in Fig. 2, each CT-SDEA unit has opposite endplates 20, 22 and connection points 24, 26 for positive and negative terminals connected to a power supply (not shown). Exemplary MASM actuators comprise a plurality of DEA units assembled in series (FIG. 3A) and/or in parallel (FIG. 3B). FIGS. 3A and 3C depict a 1x4 (one set 30 of four DEA units 32, 34, 36, 38 in series) assembly and FIG. 3B depicts a 3x4 (three parallel sets 30A, 30B, 30C of four DEA units in series) assembly. Each DEA unit is ultimately connected to a controller 39 (schematically depicted in FIG. 3A, but omitted in FIGF. 3B, to reduce clutter) via one or more electrical connectors 31.

In MASM actuator 30, DEA unit 32 is a proximal DEA unit and DEA unit 38 is a distal DEA unit. In use, the proximal DEA unit is anchored to a proximal anchor point relative to the user, which may be a frame in an exoskeleton application, wherein the frame is removably affixed to a patient. The anchoring to the frame may comprise, for example, the end plate of the DEA unit being affixed (e.g. with a fastener 33, such as a nut 33B and bolt 33A as depicted in FIG. 3C) to a proximal connector 35, which connector may be a single member or an assembly of components, and may be configured to be removably attached to the frame or other element relative to the patient. In the MASM actuators 30A, 30B, 30C, each of the three sets of four DEA units arranged in parallel has a proximal DEA unit 32A, 32B, 32C that connected to connector 35A.

Different configurations of the artificial skeletal muscle show linear additive behavior of force and displacement when the DEAs are configured in parallel and in series. When DEAs are added in parallel, the maximum force generated is multiplied by the number of actuators used (contraction displacement remains the same), while when added in series, contraction displacement is multiplied by the number of actuators used (force generation remains the same). For example, adding CT-SDEA units in series (e.g. lx l to 1 x5) shows constant force generation at ~10 N of maximum force. Configurations with two CT-SDEA units in parallel (2x 1 and 2x2) and three CT-SDEA units in parallel (3x 1 and 3x4) showed ~20 N and ~30 N of maximum force respectively. Increasing the number of CT-SDEA units increases the MASM assembly's maximum shortening almost linearly. For example, under 40 g of tensile load, a single CT-SDEA unit showed 1.3 mm of maximum shortening and an assembly of five CT-SDEA units in series showed 6.53 mm (i.e., 5 x 1.3). This linearity may decrease with increasing tensile load. As depicted in FIG. 3A, the distal DEA unit 38 is connected to a proximal end of a tension member 37. Tension member may comprise a flexible, non- extendable cable, such as a wire cable. The tension member may be a tension-only cable, such as a traditional cable, or a push-pull cable, which may offer the advantage of reducing slack in the cable and more affirmatively creating the desired displacement when the MASM expands. The tension member is not limited to any particular type of member or materials of construction; however, the non-extensibility of the cable is relatively important so that any displacement of the artificial muscle is transmitted to the desired endpoint with minimal loss. The connection between the distal DEA unit 38 and the tension member 37A is preferably via a distal connector or connector assembly 50 that provides an adjustable connection that permits adjustment of the length of the tension member. Likewise, as depicted in FIG. 3B, all three distal DEA units 38A, 38B, 38C are connected to common distal plate 51 that is connected to connector 50A, and the proximal end of tension member 37A is connected to connector 50A, preferably by an adjustable connection.

In preferred embodiments, the artificial muscle assembly may preferably be confined in a low-friction, flexible, electrically insulating covering 300, as shown in FIG. 3D, such as comprising an elastic silicone film. In addition to facilitating movement with minimal friction, the electrically insulating covering provides protection against electrical shocks, as the artificial muscle may generate strong, painful sparks, if touched. Film 300 is depicted with a clear section 305 for illustrative purposes only, to enable visualization of the underlying CT-SDEA units 32, 34, 36, 38 and connecting fasteners 33 within the covering 300. The covering 300 preferably covers all of what would otherwise be exposed surfaces of the CT-SDEA units, and fully extends between proximal connector 35 and distal connector assembly 50. FIG. 3C depicts the CT-SDEA units and connectors without the covering.

Exemplary components of a connector assembly are depicted in FIGS 4 and 5. Clamping units 40, 42 shown in FIG. 4. Each clamping unit 40, 42 comprises a pair of wire clamps 43A, 43B and 43C, 43D, respectively. Clamps 43A, 43C and 43D are depicted in a closed position, whereas clamp 43B is depicted in an open position. Clamps may be disposed in a housing such as housing 55, shown in Fig. 5. While housing 55 has certain features tailored to a specific application, as discussed in detail herein later, it is representative of exemplary housings is sized to receive the clamping units permanently affixed thereto, and has features for anchoring the housing to a frame or the like. In one embodiment, the wire clamp is placed in the open positon, the proximal end of the tension member is threaded into the clamp as far as desired to give the tension member a desired length, and the clamp is closed. This permits adjustability of the length of the tension member to tailor the MASM assembly to a particular patient.

The distal end of the tension member 37 is typically affixed to a distal anchoring point, such as to a footplate 150 as depicted in FIGS. 14A and 14B, and discussed in detail herein later. The distal end may be affixed with a non-adjustable connection, with the proximal connection providing ample modularity and adjustability, or the distal end may be also affixed with an adjustable connection. In some embodiments, the tension member 37 may form a loop in which both free ends are proximal ends that connect to the proximal anchoring point (e.g., one each in clamps 43A and 43B, and the distal end of the tension member is the distalmost point of the loop, where that loop connects to the distal anchor point.

With the MASM actuators thus anchored at their proximal end and connected at their distal end to a distal anchoring point via a tension member, the controller, which is attached to a power source and to the plurality of dielectric elastomer units, applies a voltage to the plurality of dielectric elastomer units in each MASM assembly simultaneously, thus causing contraction or expansion of plurality of dielectric elastomer units, which contraction or expansion causes displacement of the tension member.

Electronics

A high voltage (HV) amplifier (such as, but not limited to, an RC250- 1.5P,Matsusada Precision Inc., Shiga, Japan, with 1500 V and 165 mA maximum output voltage and current respectively) may be used for providing driving voltage in a range of 0-1230 V for the CT-SDEAs. FIG. 6A depicts the electrical circuitry of the HV amplifier, VDC, discharge resistance, Rds, and CT-SDEA. Although depicted as a CT-SDEA, similar circuitry may be applicable to other types of DEA as well. The DEA may be modeled as a parallel variable capacitor, C, and resistor, R, in series with two electrode resistors, RE. The combination of RE and R model the non-ideal leakage current through the dielectric. Variability in the C, R, and RE is due to length modulation of the DEA, and consequently, changes in the electrical characteristics of the compliant capacitor. Electrode resistance of each of the CT-SDEA's layers has been reported at 50 kQ. Considering 1600 parallel layers of DE material in each CT-SDEA unit, the total electrode resistance may be as small as 31.25 W in a resting state. A discharge resistor Rds (e.g. 1 MW) may be connected in parallel to increase the rate of discharging the SDEA, and thereby, increase the pace of recovery, i.e., returning to initial state.

To charge the CT-SDEA for full contraction, the transient current consumption of one actuator may reach as high as 22.77 mA for a couple of milliseconds. For safety concerns, current may be limited by connecting a resistor in- series with the actuator, which also limits the contraction velocity.

In active exoskeletons, where assisting motion is usually the main goal, the shortening, i.e., longitudinal displacement, of the artificial skeletal muscle is designed to be large enough to allow for its contraction alongside the impaired muscle. Because the maximum strain of the DEAs is substantially lower than skeletal muscle, it may be advantageous to leverage the substantial force of the SDEAs and magnify the displacement using a mechanical linkage, or to assist the user's skeletal muscle contraction only in a portion of the user's range of motion (RoM).

Self-Sensing

Unlike conventional actuators, DEA units do not require an independent sensing device such as encoder or tachometer. Changes in DEA unit electrical characteristic may be used to infer length modulation. When multiple degrees of freedom are needed, in applications such as rehabilitation robotics, this feature may substantially reduce the cost, efficiency and complexity of the system.

Sensor-free closed-loop control of a DEA unit or assembly of a plurality of DEA units is attainable using a self-sensing length modulation system that mimics the proprioception in human motor control. During actuation, the changes in the electrical characteristics of the DEA, as a compliant capacitor, are used for sensing the

longitudinal contraction of the actuator. Self-sensing may be effected by using a low- pass filter designed by connecting a resistor in series with the DEA, then superimposing a high frequency signal to the driving, low frequency, activation waveform, such as is depicted in FIG. 8. Measuring the amplitude of the output of the low-pass filter permits displacement of the DEA to be inferred. For example, changes in the amplitude of the 10 Hz sinusoid signal at resting length (0%), 5% compression, and 5% extension, are depicted in FIG. 9. As the DEA is compressed, the amplitude of the sensing signal decreases, and as it extends, the signal's amplitude increases. The self-sensing information is then fitted to actual displacement information to establish the

relationship between self-sensed length and actual displacement or degrees of joint rotation. Real-time length self-sensing of the CT-SDEA in one embodiment positioned relative to displacement sensor data is depicted in FIG. 10. The self-sensing data may be derived from the changes in the peak-to-peak (P-P) value of Voutput (as depicted in FIG. 10, the self-sensing data shown as a negative, to follow the trend of the output laser displacement sensor). The equation of the fitted curve depicted in FIG. 10 for the exemplary CT-SDEA was found to be.

0 82?.r* 8 S43& 4 ODOG425 It should be understood, however, that the invention is not limited to any particular correlation between the self-sensed data and actual displacement. Importantly, the self-sending data corresponding to a particular arrangement of DEA units may be mapped to a curve that permits use of a single self-sensing relationship for a single measured DEA unit that can be mapped to the MASM as a whole (comprising, e.g.,

1x4, 3x5 DEA units), rather than sensing information for each separate DEA in the stack or for the stack as a whole.

Many self-sensing techniques are known and have been documented, and the invention is not limited to any particular self-sensing technique. However, the exemplary DEA self-sensing mechanism as described above is now described in more detail, without limitation thereto. This DEA self-sensing mechanism works by inferring longitudinal contraction information from changes in the electrical characteristics of the DEA as illustrated in FIG. 11. The impedance of the DEA ( _c) changes by its

longitudinal contraction, which can measure by recording Voutput, using Equation 1. DAE's impedance ( X_c ) is inversely proportional to the DEA's capacitance, C. By using Equation 2, and the known frequency of Vmput, f, C can be measured. Finally, using Equation 3, the thickness of the DEA {d) can be calculated.

As the DEA as depicted in Figure 11 contracts longitudinally, i.e., d decreases, its capacitance increases (per Equation 3); consequently, in the frequency domain, the cut-off frequency, F_c decreases (per Equation 4) :

where R is the total resistor in the electrical circuitry including R_s.

In other words, when a DEA longitudinally contracts its capacitance increases, thus the F_c decreases, consequently, the attenuation of input signal increases, which results in Voutput decrease, as depicted in FIG. 7.

The electrical model of a DEA depicted in FIG. 6A, in which C_P and R_P are its variable capacitor and resistor, respectively, can be simplified to the model shown in FIG. 6B, because the variability in Cp, R, and RE are due to the length modulation of the CT-SDEA, but in the low frequency range (e.g., 10-100 Hz), RE can be neglected.

Using Laplace transformation, the impedance of the CT-SDEA can be calculated using Equation 5.

By adding Equation 5 to Equation 1, the V output to Vm_Put transfer function can be derived (Equation 6), which is a first order lag system, high pass filter.

For a stacked DEA, C and R_P can be calculated using the slightly modified formulas Equations 7 and 8, respectively. Where n is the number of stacked layers in the CT-SDEA; A_e is the active area of each layer (the area covered by electrode); o is the resting length of the actuator excluding the thickness of its end plates; £_rand £o are relative permittivity and void permittivity respectively of the polymer (silicon rubber for CT-SDEA); V is activation voltage; and d is length of the activated actuator.

Using Equations 6 and 7, the polymer's time constant T_p, can be calculated. Equation 9 shows that the actuator's time constant only depends on the polymer's physical characteristics:

In Equation 9, /% is the resistivity of the dielectric material, i.e., silicon rubber in CT-SDEA.

Table I provides definitions for R_p and C_P parameters and exemplary measured or reported values.

TABLE I

The values for R_Po and s may be calculated empirically. During an isometric contraction, the length of the CT-SDEAs was held constant at a rest length under no compression or extension. After application of a square activation waveform activation, with amplitude of 1230 V and period of 5 s, the CT-SDEAs showed 38 ms time constant, during the discharge period. Using Equation 8 and considering a 1 MW discharge resistance connected in parallel with the CT-SDEA, R_Po and Ps were calculated at 191 ^c 10⁵ W and 1.8264x lOⁿ, respectively.

The gain of a first order lag system, in time domain, K (Equation 10), for a sinusoidal input with frequency of f, is presented in Equation 11. This gain may be calculated for different Rs and input frequencies as the length of the CT-SDEA changes. The time constant, t and A in H(s) (Equation 6) is shown in Equation 12 and 13, respectively.

Self-sensing circuits may be connected to the controller for the MASM to provide feedback to the controller, and the controller may thus be configured to control contraction or expansion of each MASM assembly of DEA units based upon the detected longitudinal state. Use of such self-sensing may permit the controller to operate as a PID (proportional-integral-derivative) controller. In rehabilitation robotics, closed-loop controls (such as provided by a PID controller) are more preferable due to reliability. In such applications, precise displacement feedback may be an integral component of a robust closed-loop motion control.

EMG control

In some embodiments, however, open-loop control of DEAs using biosignals, i.e., EMG, EEG, and ECG, for wearable technology may also be used. Thus, the MASM actuator may comprise an electromyographic (EMG) signal processing interface connected to the controller and to one or more EMG sensors configured to detect electrical activity of a living muscle, wherein the controller is configured to control contraction or expansion of the plurality of DEA units based upon a signal from the EMG signal processing interface. For example, FIG. 15C schematically illustrates a sensor 10 connected to a processor 12 connected to a controller 39, which general depiction may be relevant to multiple, different embodiments. With respect to an EMG- controlled embodiment, sensor 10 may comprise an EMG sensor, and processor 12 may comprise a signal processing interface connected to controller 39, with the controller configured to trigger MASM actuators (disposed in pockets 1580, 1590).

Exoskeletons

Figure 12A depicts a CAD model of the schematic components of an exemplary exoskeleton 1200, powered by a 3x4 MASM 1202. The design contains three brackets located around the end plates of connecting CT-SDEA units to confine undesirable motion of the artificial muscle, and minimize any impact caused by bowing. The artificial muscle is preferably confined in a low friction, flexible, electrically insulating covering, as shown in FIG. 3D. To further restrict undesired motion, the artificial muscle assembly may be also be attached to a harder skeletal-like connecting frame, such as comprising top plate 1206 which connect the proximal CT-SDEA units to the proximal portion 1212 of the exoskeleton frame, and bottom plate 1208 connecting the distal CT-SDEA units to the tension member 1210, which is connected to a proximal end of distal portion 1214 of the exoskeleton frame. In the embodiment depicted in Fig. 12A, proximal portion 1212 of the frame is configured to be anchored about a shoulder of the user and the frame distal portion 1214 comprises a cuff configured to be anchored to a forearm of the user. The mechanisms for anchoring the frame to the user are only shown schematically, and it should be understood that the frame may have any geometry, including to provide a desired degree of rigidity for the proximal (shoulder) portion, additional components that, for example, are disposed across the user's body to the other side (e.g. across the rib cage, or to the other shoulder).

Adding a mechanical transmission, such as a mechanical linkage, for magnifying displacement may be highly desirable for some exoskeleton applications.

An exemplary magnification system, for use in connection with an elbow exoskeleton, is depicted in FIG. 12B as a cam-shaped pulley system. As depicted, proximal end of tension member 1220 (e.g. wire cable) is adjustably affixed to connector housing 1222 attached to the distal CT-SDEA unit 1224 of MASM assembly 1226. Member 1220 extends around pulleys 1230, 1232 and connects to proximal side of cam 1240, which is rotatable about axial pin 1242. The distal side of cam 1240 is attached to another tension member 1244, which attaches to an anchor point 1246 on the distal portion 1248 of the exoskeleton frame. In this way, a first amount of displacement of the tension member 1220 is magnified by the foregoing system into a second amount of displacement of the distal portion of the frame relative to the proximal portion that is greater than the amount of displacement of tension member 1220. The invention is not limited to any particular type of magnification assembly.

Exemplary exoskeleton embodiments for use about an ankle joint of a user are depicted in FIGS. 13A-15D. Like the exoskeleton embodiment depicted in FIGS. 12A and 12B for use about an elbow joint of a user, each embodiment comprises a frame comprising at least an proximal portion configured to be removably attached to a user proximal a joint of the user and a distal portion configured to be removably attached to the user distal of the joint of the user, with the proximal and distal portions configured to move relative to one another with at least one degree of freedom. One or more modular MASM actuators, each anchored by an actuator proximal anchoring member attached to the proximal portion of the frame at the proximal anchor point, include a tension member having its distal end attached to the distal portion of the frame at a distal anchor point. Each tension member preferably has an adjustable length between the proximal anchor point and the distal anchor point. In all of the embodiments, displacement of the respective tension member caused by contraction or expansion of each MASM actuator causes corresponding displacement of the distal portion of the frame relative to the proximal portion of the frame.

As shown in the embodiment 1300 depicted in FIGS. 13A - 13D, the proximal portion 1302 of the exoskeleton frame comprises a proximal cuff 1304 configured to be mounted proximal of a knee of the user and a distal cuff 1306 configured to be mounted distal of the knee of the user, with the proximal and distal portions connected to one another on opposite sides of the knee of the user with opposite exoskeletal knee joints 1308, 1310. The distal portion 1320 of the frame comprises a footplate configured for positioning beneath a foot of the user, such as the footplate depicted in FIGS. 14A and 14B. In some embodiments, the footplate may extend above the plantar aspect of the user's foot, and may even form a shape in the nature of a shoe. The depiction in FIGS. 13A-13D is intended only to be schematic, and tends to show connections directly to the shoe, but it should be understood that the connections may all terminate in a footplate similar to that shown in FIGS. 14A and 14B. The term "footplate," however, is not intended to refer only to the design shown in FIGS. 14A and 14B, and encompasses any distal portion configured with at least one component that extends beneath the user's foot, regardless of whether other elements may be present that extend above the sole of the user's foot. The system includes ankle dorsiflexor MASM actuator 1330 configured to augment ankle dorsiflexor muscle function and ankle plantarflexor MASM actuator 1340 configured to augment ankle plantarflexor muscle function. An inversion MASM actuator 1350 is configured to augment ankle inversion muscle function (function that causes the sole of the foot to point in a medial direction) and an eversion MASM actuator 1360 configured to augment ankle eversion muscle function (function that is the opposite of inversion). Each of actuators 1330, 1360, and 1350 are depicted as 1x4 MASMs. Actuator 1340 is depicted as a 2x4 MASM. The invention is not limited to any particular configuration or number of units for each muscle, however, certain muscles may require more force in comparison to others, because of physiological needs. Because the plantarflexors create the necessary propulsion for walking, they typically generate more force in comparison to the other muscles around the ankle, and their artificial counterparts therefore also benefit from being able to generate proportionately relative forces. In fact, one advantage of the invention is the modular capability to tailor the muscle travel and force to the individual user. For example, a smaller or younger subject may not need the displacement of a 1x4 MASM (or may not have the physiological leg size to support the length of a 1x4 MASM), and may be able to use a 1x3 MASM in one or more locations. As that individual grows, he or she may need a larger unit, and an additional DEA unit may be added to the stack when needed (with all appropriate electrical and physical connections). Embodiments of the DEA units may be provided in any shape or size and with suitable connection interfaces to facilitate such modularized tailoring of the structures. The location of the connection point of the tension members to the distal portion of the exoskeleton (e.g. the footplate) may be adjusted based on the gait deviation and foot deformity of the user, which may be a highly useful feature because of the heterogeneity of cerebral palsy and ankle control dysfunction.

The tension member 1332 attached to actuator 1330 is attached to a position on top of the user's foot, as depicted in FIG. 13A to an area near the toecap 1334 at multiple (four) attachment points, after branching into four branches. In other embodiments, 1332 may be connected to strap 1370 such as via a harness such housing 55, depicted in FIG. 5, as described further herein. In still other embodiments, tension member 1332 may not branch at all, or may branch into fewer or more than four branches. Tension member 1352 is attached to an anterior, medial location of the footplate. Tension member 1362 branches into a first branch 1362A attached to an anterior, lateral location of the footplate and a second branch 1362P attached to a posterior, lateral location of the footplate. Tension member 1342 is attached to a posterior, central location of the footplate. Footplate 150 depicted in FIGS. 14A and 14B depicts a plurality of holes for receiving tension members. Hole 142 receives tension member 1352 attached to inversion MASM actuator 1350. Hole 144 receives tension member 1362 attached to eversion MASM actuator 1360. Heel support 146 comprises one or more holes for receiving tension member 1342 attached to plantarflexor MASM actuator 1340. Slots 148 on opposite sides of the footplate are for receiving a strap (e.g. 1370 shown in FIG. 13B) that extends across the instep of the shoe. Holes 152 on opposite sides of the footplate positioned outward of slots 148 are for receiving a plurality of wire members for reinforcing the strap. Strap 1370 thus is disposed between the wire members and the top of the wearer's foot, also protecting the wearer from the cables rubbing against the wearer's skin.

Referring now to FIG. 5, housing 55 is depicted as an exemplary design for mounting on the instep strap for anchoring tension member 1332, connected to dorsiflexor MASM actuator 1330, to the footplate. Housing 55 includes a slot 52 for receiving the instep strap (e.g. 1370) and a seat 54 for receiving a one or more clamping units (e.g. 40, 42) comprising one or more clamps (e.g., 43A, 43B, 43C). In view of the three holes 152 depicted on footplate 150, a suitable clamping unit for use in connection with housing 55 for use on the instep may comprise a clamping unit having at least three clamps, each clamp configured to receive one or more wires from either or both sides of the footplate. Wires may be mounted to each hole on the footplate such that each clamp receives a wire from both sides, or six clamps may be provided with each receiving only one wire from one side. Wires may be mounted to all or fewer of the holes on each side, and more or fewer holes may be provided for mounting the wires. The wires provide adjustability and firmness for transmitting the forces to the footplate from the housing, whereas the strap provides cushioning for the instep of the wearer. Hole 56 receives the distal end of tension member 1332, thus connecting dorsiflexor MASM actuator 1330 to the footplate via the cables adjustably secured within housing 55.

Slot 154 is configured to receive the distal end of a connecting member for another powered exoskeleton embodiment for mounting about an ankle depicted in FIGS. 15A-15D. Unlike the embodiment depicted in FIGS. 13A-13D, in which the distal and proximal portions of the exoskeleton are not connected to one another by frame elements, the embodiment depicted in FIGS. 15A-15D includes an exoskeletal connector 1502 that extends between the distal portion 1504 of the frame (e.g.

footplate 150) and the proximal portion 1506 of the frame. The exoskeletal connector 1502 is affixed to a lateral connection point on the footplate (e.g. slot 154) and to a posterior, medial connection point 1510 on the proximal portion of the frame. Connector 1502 extends between the connection points posterior to the lower leg of the user, as best depicted in FIG. 15D. Although shown in a particular configuration in FIGS. 15A-15D, it should be understood that the connection points for the exoskeletal connector to the footplate and to the proximal portion of the frame are not limited to any particular locations or relative locations (e.g. medial, lateral, posterior, anterior) and the connector may extend along any portion of the lower leg of the user.

In the embodiments depicted in FIGS. 15A-15D, each of the MASM actuator assemblies are configured to move freely in a pocket 1580, 1590 formed in or attached to a textile member 1520 (e.g. a sock), and are configured to be removably anchored at their proximal ends to the proximal portion of the frame. For example, connection mechanisms 1530 located where the MASM actuator assemblies are anchored to the proximal portion 1506 of the exoskeleton may comprise mating interlocking (e.g. male/female) connectors, which may be held in place by any means known in the art, including but not limited to the use of biased tabs and slots that create a snap-fit, magnetic connectors, or the like. As shown in FIGS. 15A-15D, the sock extends above the user's knee. Ideally, however, the connections can be snapped into place after the user threads his or her leg through the sock. Preferably, each respective tension member is confined for some portion of its length within relatively low friction conduits. The low friction conduit may comprise a tube, such as comprised of ultrahigh molecular weight (UHMW) polyethylene or other lining, that provides low friction to allow the cables to move up and down freely, but with sufficient side-to-side anchoring to minimize or prevent side-to-side displacement and rubbing against the wearer's skin. Cuff 1540, disposed over the tubes, minimizes bowing of the tension members. Instead of cuff 1540, the wire conduits may be attached to (e.g. sewed into) the sock, and may be less visibly defined than as depicted in the figures (for emphasis). The low friction conduits may extend nearly the entire length of the tension members between connection points to the MASM and the exoskeleton, and may be individually formed within the textile sock rather than part of a cuff.

Housings 1536, 1546, 1556, 1566, similar to 1222 referenced in FIG.

12B, are located at the distal end of each MASM (hidden within pockets 1580, 1590, but similarly positioned to MASMs 1330, 1340, 1350 and 1360, respectively, as depicted in FIGS. 13A-13D). Each housings holds one or more clamping units (e.g. 40, 42) with at least one clamp (e.g. 43A) that permits adjustment of the length (and tension) of the respective tension members (1532, 1542, 1552, 1562) at their respective proximal ends. The distal ends of the tension members are typically non- adjustably fixed to the footplate.

EXOSKELETONS FOR GAIT REHABILITATION The MASM-powered exoskeletons as described herein may be useful for any type of use, including rehabilitative or therapeutic uses for normally healthy subjects recuperating from injuries, and for any type of individual (adults or children), but particularly useful embodiment may include frames configured for removable attachment to a pediatric user, and in particular, for use by children with CP. In one embodiment, the exoskeletons may be controlled using gait phase detection (GPD) for triggering functional electrical stimulation (FES) in rehabilitation systems.

Typical gait has been described as a series of seven contiguous phases: Loading Response (LR), Mid-Stance (MSt), Terminal Stance (TSt), Pre-Swing (PSw), Initial Swing (ISw), Mid-Swing (MSw), and Terminal Swing (TSw). Methods for detecting these phases during walking, known as gait phase detection (GPD), are known and have been well documented with respect to healthy subjects. See, e.g., Behboodi, A. et al., "Seven phases of gait detected in real-time using shank attached gyroscopes", Proceedings of the Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS; Milan, Italy, 2015; Vol. 2015- November, pp. 5529-5532, incorporated herein by reference. The foregoing method comprises a simple two-gyroscope GPD system that detects all seven phases of gait.

Exemplary GPD systems for identifying specific gait events for triggering FES to enhance gait in children with CP benefit from the use of a modified algorithm as compared to that previously described. For healthy adults and typically developing (TD) children, medio-lateral shank angular velocity (*¼; ) has a definitive pattern 1610 during the gait cycle, as depicted in FIGS. 16A and 16B. The typical pattern comprises three positive peaks 1612, 1613, 1614 in the stance phase followed by a deep negative peak 1615 in the swing phase. This pattern is typically detected by attaching an inertial measurement unit (IMU), such as a gyroscope that measures angular velocity, to the shank of the subject. An exemplary IMU 10 suitable for this purpose may include an IMU made by APDM Inc., Portland, OR, USA, contain three-axis accelerometers, gyroscopes and magnetometers, and is depicted schematically as sensor 10 in FIG. 15C (but omitted in other figures to reduce clutter). Although some IMU devices may comprise a combination of accelerometers, gyroscopes and magnetometers measuring all relevant degrees of freedom, unless expressly stated, the term "IMU" as used herein does not imply a specific combination of sensors, and may refer to units comprising only an accelerometer, only a gyroscope, or only a magnetometer (or some

combination of two of the foregoing), and may also refer to units configured to measure in only a single degree of freedom or in any combination of degrees of freedom fewer than all of those relevant in the measured system. One IMU (or gyroscope) is preferably attached to the lateral side of each shank of the subject with the Z-axis of the IMU aligned in the medio-lateral direction (i.e., axis of rotation of the knee in the sagittal plane), and the z-component of the gyroscope data (w™;) is used as an input to the GPD system (e.g. controller 39). The IMUs may be aligned such that knee flexion and extension resulted in respective positive and negative values of o)mi.

The GDP-TD algorithm (used for typically developing children) was tested on children with CP. Although children with CP do not often exhibit typical gait events (e.g., those with equinus gait may lack heel strike), shank angular velocity shows similar features and can still be used to determine gait phases. However, some modifications to the GPD algorithm were found to be beneficial. In particular, while w™; typically has easily identifiable peaks and zero-crossings, the lack of a distinct peak at toe-off/end-contact (TO/EC) confounded ISw detection for children with CP. This issue was mitigated by using the arithmetic sum of all three components of shank angular velocity (a)_SUm) instead of « £ to detect the TO/EC peak. The summed signal featured a more prominent peak at TO/EC, isolating it from spurious peaks present in the &½ signal. Because o)_Sum slightly leads ami in time and because the MSw zero-crossing closely follows TO/EC, o)_Sum was also used to detect the MSw zero crossing. This reduced the chances of erroneously detecting MSw (in w™;) before detecting ISw (in

(Osum) .

The foregoing is now illustrated with respect to FIGS. 16A and 16B.

Lines 1610 in FIGS. 16A and 16B depicts shank angular velocity about the medio- lateral axis (ami) for typically developing children and healthy adults during treadmill walking. Gait phase onset is based on the indicated peaks and zero-crossings. Four gait phase events are detected using ipsilateral shank angular velocity (loading response (LR), initial swing (ISw), mid-swing (MSw) and terminal swing (TSw). The three remaining gait phase events are detected using contralateral shank angular velocity (mid-stance (MSt), terminal stance (TSt) and pre-swing (PSw)). The onset of LR corresponds to initial contact (IC)/heel strike (HS) and the onset of ISw corresponds to toe-off/end-contact (TO/EC). IC may also be considered a separate phase from LR onset. The gait phase detection system as referenced herein is capable of splitting LR into an IC phase and LR. In such configurations, the IC phase may be identified as the current LR onset (zero-crossing) to the first positive peak 1612. FIG. 16B illustrates a comparison of line 1610, which depicts representative shank angular velocity about the medio-lateral axis for a typically developing (TD) child (top), to line 1620, which depicts representative shank angular velocity about the medio-lateral axis for a child with cerebral palsy (CP). A distinct peak 1613 is visible at the onset of ISw (TO/EC) in TD subjects while the peak 1623 is less distinct in CP subjects. The sum of all three components of shank angular velocity ( w_SUm ) in line 1630 shows a more distinct peak at ISw onset 1633, and was used for ISw detection in children with CP.

Even with the increased detection reliability of ISw with o)sum, extraneous peaks and zero-crossings due to spasticity may result in false detections of ISw and MSw. To mitigate this, the following criteria were found to be beneficial:

(1) ISw detection was blocked until at least 60% of the average gait cycle duration had elapsed since the last LR detection,

(2) MSw detection was blocked until at least of 25% of the average of the last 10 gait cycle durations had elapsed since ISw; and

(3) peak detection threshold values were set to 25% of the smallest ISw peak height and highest TSw valley depth observed over the first few previous gait cycles.

Thus, controller 39 may comprise a finite state control system including one or more feedback sensors 10 (e.g. which, for purposes of describing this embodiment, may be an IMU mounted on the lateral side of each exoskeleton for each leg) configured to activate a preset sequences of the plurality of MASM actuators (disposed in pockets 1580, 1590) in response to specific detected system state conditions. As depicted in Fig. 15C, pocket 1590 may comprise a plurality of actuators, including actuators for dorsiflexion, eversion, and eversion. In particular, data from IMU sensor 10, processed by processor 12 (which may be separate from or integral with controller 39, may be configured to detect different phases of ankle motion during a walking gait of the user, and the controller may further comprise plurality of stimulators for activating preset sequences of the plantarflexor, dorsiflexor, inversion, and eversion MASM actuators in synchrony with the detected phases of the walking gait, as illustrated schematically in the block diagram 1700 shown in FIG. 17. As depicted in FIG. 17, a subject 1702 walks on a treadmill 1704 with motion sensors 1710, 1720 attached to each of the left and right shanks. The motion sensors send shank angular velocity measurements to a control computer 1730. A gait phase detection algorithm 1732 detects gait phase transitions based on shank angular velocity. Upon gait phase transition, a stimulation control algorithm 1734 sends stimulation commands (indicating current amplitude and pulse duration) to left and right stimulators 1740, 1750 based on the subject's stimulation protocol. Systems without inversion, and eversion MASM actuators may similarly activate only

plantarflexor and dorsiflexor MASM actuators. In preferred embodiments, IMU sensor 10 comprises a gyroscope configured to detect bilateral shank angular velocity of the user for determining gait phase transitions. For powered exoskeletons configured for therapeutic use in connection with pediatric subjects with cerebral palsy, the controller may be configured with an algorithm for detecting gait phase based upon bilateral shank angular velocity about the z-axis, or based upon a sum of the bilateral shank angular velocity about all three axes of the gyroscope (depending upon the degree of CP and/or phase being detected), and the GPD detection algorithm in the processor may include detection criteria tailored to gait phase transitions of children with cerebral palsy. As described above, the detection criteria tailored to gait phase transitions of children with cerebral palsy may be generally described as including a delay in Initial Swing (ISw) detection until after a predetermined percentage of elapsed gait cycle, a delay in Mid-Swing (MSw) detection until a predetermined number of samples after ISw, and a peak detection threshold for ISw set to a predetermined percentage of a smallest detected peak in a predetermined number of cycles, with a peak detection threshold for Terminal Swing (TSw) set to a predetermined percentage of a smallest detected valley in a predetermined number of cycles. Some of the phases may only need to use angular velocity about the z-axis, whereas ISw and MSw may need to use summation of angular velocities, depending on the severity of the cerebral palsy of the individual user.

One distinctive aspect of MASM actuators is the elastomeric nature that closely mimics the passive elastic component of biological muscles. This can be modulated by the elastic silicone cover with various thickness and stiffness. For example, during TSw, the platarflexors may produce a significant amount of passive force by contracting eccentrically, which force is released during the push off phase PSw. This helps in generating a substantial amount of propulsive force for pushing the center of mass of the body forward during walking gait.

Although described with respect to specific types of exoskeleton embodiments, it should be understood that the invention is not limited to any particular uses for the MASM actuators as described herein. Furthermore, although emphasized for use in connection with elbow and ankle joints, the exoskeletons using MASM actuators may have utility for use in connection with any joint. Although discussed herein with respect to therapeutic or rehabilitative uses, exoskeletons as described herein are not limited to any particular uses.

Although the invention is illustrated and described herein with reference to specific embodiments, the invention is not intended to be limited to the details shown. Rather, various modifications may be made in the details within the scope and range of equivalents of the claims and without departing from the invention.

Claims

What is Claimed :

1. A modular artificial skeletal muscle (MASM) actuator, the MASM actuator comprising :

a plurality of dielectric elastomer units assembled in series or in parallel; a proximal anchoring member connected to a proximal end of at least one proximal dielectric elastomer unit and configured for attachment to a proximal anchor point relative to a user;

a tension member having a proximal end attached to a distal end of at least one distal dielectric elastomer unit and a distal end configured for attachment to a distal anchor point relative to the user; and

a controller attached to a power source and to the plurality of dielectric elastomer units, the controller configured to cause simultaneous contraction or expansion of plurality of dielectric elastomer units, which contraction or expansion causes displacement of the tension member.

2. The MASM actuator of claim 1, further comprising a flexible, low friction covering disposed over an external surface of the plurality of dielectric elastomer units.

3. The MASM actuator of claim 1, further comprising a self-sensing circuit connected to the controller, the self-sensing circuit configured to detect a relative longitudinal state of at least one sensed dielectric elastomer unit in the MASM and to provide feedback to the controller, wherein the controller is configured to control contraction or expansion of the plurality of dielectric elastomer units based upon the detected relative longitudinal state of the at least one sensed dielectric elastomer unit.

4. The MASM actuator of claim 3, wherein the controller comprises a

PID controller.

5. The MASM actuator of claim 1, further comprising an electromyographic (EMG) signal processing interface connected to the controller and to one or more EMG sensors configured to detect electrical activity of a living muscle, wherein the controller is configured to control contraction or expansion of the plurality of dielectric elastomer units based upon a signal from the EMG signal processing interface.

6. The powered exoskeleton of claim 1, wherein the tension member comprises a flexible, non-extendable cable.

7. A powered exoskeleton configured to be worn by a user, the powered exoskeleton comprising:

a frame comprising at least an proximal portion configured to be removably attached to the user proximal a joint of the user, a distal portion configured to be removably attached to the user distal of the joint of the user, the proximal and distal portions configured to move relative to one another with at least one degree of freedom;

one or more modular MASM actuators of claim 1, each MASM actuator proximal anchoring member attached to the proximal portion of the frame at the proximal anchor point, each MASM actuator tension member distal end attached to the distal portion of the frame at the distal anchor point, each tension member having an adjustable length between the proximal anchor point and the distal anchor point;

wherein the displacement of the respective tension member caused by contraction or expansion of each MASM actuator causes corresponding displacement of the distal portion of the frame relative to the proximal portion of the frame.

8. The powered exoskeleton of claim 7, wherein the frame is configured for placement about an elbow joint of the user.

9. The powered exoskeleton of claim 8, wherein the frame proximal portion is configured to be anchored about a shoulder of the user and the frame distal portion comprises a cuff configured to be anchored to a forearm of the user.

10. The powered exoskeleton of claim 7, wherein the at least one tension member is connected to a displacement magnification system that converts a first amount of displacement of the tension member into a greater amount of displacement of the distal portion of the frame relative to the proximal portion.

11. The powered exoskeleton of claim 7, wherein the frame is configured for removable attachment to a pediatric user.

12. The powered exoskeleton of claim 7, wherein the frame is configured for placement about an ankle joint of the user.

13. The powered exoskeleton of claim 7, comprising a plurality of MASM actuators.

14. The powered exoskeleton of claim 13, wherein the controller comprises a control system including one or more feedback sensors and is configured to activate preset sequences of the plurality of MASM actuators in response to specific detected system state conditions.

15. The powered exoskeleton of claim 13, wherein the frame is configured for placement about an ankle joint of the user, and the system includes at least one plantarflexor MASM actuator configured to augment ankle plantarflexor muscle function and at least one dorsiflexor MASM actuator configured to augment ankle dorsiflexor muscle function.

16. The powered exoskeleton of claim 15, further comprising at least one inversion MASM actuator configured to augment ankle inversion muscle function and at least one eversion MASM actuator configured to augment ankle eversion muscle function.

17. The powered exoskeleton of claim 16, wherein the controller comprises a control system including a sensor configured to detect different phases of ankle motion during a walking gait of the user and to activate preset sequences of the plantarflexor, dorsiflexor, inversion, and eversion MASM actuators in synchrony with the detected phases of the walking gait.

18. The powered exoskeleton of claim 17, wherein the sensor configured to detect different phases of ankle motion during the walking gait comprises a gyroscope configured to detect bilateral shank angular velocity of the user for determining gait phase transitions and a plurality of stimulators for activating the MASM actuators.

19. The powered exoskeleton of claim 18, wherein the powered exoskeleton is configured to rehabilitate a gait of a pediatric user with cerebral palsy, and the controller is configured with an algorithm for detecting gait phase based upon z axis bilateral shank angular velocity or a sum of bilateral shank angular velocity about three axes and which includes detection criteria tailored to gait phase transitions of children with cerebral palsy.

20. The powered exoskeleton of claim 19, wherein the detection criteria tailored to gait phase transitions of children with cerebral palsy includes a delay in Initial Swing (ISw) detection until after a predetermined percentage of elapsed gait cycle, a delay in Mid-Swing (MSw) detection until a predetermined number of samples after ISw, and a peak detection threshold for ISw set to a predetermined percentage of a smallest detected peak in a predetermined number of previous cycles, and a peak detection threshold for Terminal Swing (TSw) set to a predetermined percentage of a smallest detected valley in a predetermined number of previous cycles, wherein the peak for ISw is evaluated as a sum of the bilateral shank angular velocity about the three axes.

21. The powered exoskeleton of claim 12, wherein the distal portion of the frame comprises a footplate configured for positioning beneath a foot of the user.

22. The powered exoskeleton of claim 21, wherein the proximal portion of the frame is configured for positioning about a lower leg of the user and further comprises an exoskeletal connector connecting the footplate to a posterior, connection point on the proximal portion of the frame, and configured to extend along the lower leg of the user.

23. The powered exoskeleton of claim 21, wherein the proximal portion of the frame is configured for positioning about a lower leg of the user and has no exoskeletal member connecting the footplate and the proximal portion of the frame other than the MASM actuators and tension members.

24. The powered exoskeleton of claim 12, wherein the plurality of dielectric elastomer units of each MASM actuator are disposed in a pocket that is attached to a textile member and configured to be removably anchored to the proximal portion of the frame, and each respective tension member is confined for some portion of its length within a relatively low friction conduit in the textile member.

25. The powered exoskeleton of claim 12, wherein the proximal portion of the frame comprises a first portion configured to be mounted proximal of a knee of the user, a second portion configured to be mounted distal of the knee of the user, the first and second portions connected to one another on opposite sides of the knee of the user with opposite exoskeletal knee joints.

26. The powered exoskeleton of claim 12, further comprising a dorsiflexor MASM actuator configured to augment ankle dorsiflexor muscle function, wherein the tension member connected to the dorsiflexor MASM actuator connects to a housing mounted to an adjustable, cable-reinforced instep strap connected to opposite sides of the footplate.