Design and Validation of a Two-Degree-of-Freedom Powered Ankle-Foot Orthosis with Two Pneumatic Artificial Muscles

Highlights

• The subtalar joint is an important part of balance training.

• System design and validation were proposed for a 2-df PAFO.

• Spatial formulas for the talocrural and subtalar joints were calculated and added.

• For precise control, a sliding mode control for a nonlinear system of PAM was used.

• Kinematics, workspace, frequency response, and clinical experiments were analyzed.

Abstract

Powered ankle-foot orthosis (PAFO) is a field of wearable robotics that improves the lives of people of old age or with physical impairments by aiding in the wearer's ankle joint movements. Most of the PAFOs developed thus far offer only one degree-of-freedom (dof), which uses the talocrural joint alone as the axis of rotation, where the emphasis is on moving forward. However, because this type of wearable robotics has evolved, developing PAFOs for functional rehabilitation has become necessary. This enhances the quality of walking rather than providing simple rehabilitation. The subtalar joint is responsible for the rotation of the inversion and eversion of the ankle, enabling balanced walking in humans, and is an important part of balance training for elderly people or stroke patient rehabilitation. Therefore, we developed a 2-dof PAFO that uses a calculated spatial formula of the subtalar joint based on anatomical data. In recognition of the fact that the pneumatic artificial muscle (PAM) can be used within the contracting range alone because of the instinct of the PAM, we analyzed the workspace of the fabricated PAFO through kinematic analysis and verified the possibility of using the PAFO during the gait cycle. Experiments were also conducted on the closed-loop force frequency response using a sliding mode control of the solenoid valve to validate the control characteristics. Lastly, clinical experiments with healthy subjects were conducted for validation in wearing conditions.

Introduction

The foot is in direct contact with the ground and plays the most important role in human walking. The joints involved in the movements between the foot and calf are the ankle joints, and gait can be realized through the movement of these joints [1,2]. During the gait cycle, propulsion is produced through plantar flexion at the point of push-off, and the foot drop during the swing phase is prevented by dorsiflexion. However, when aging causes damage to the muscles responsible for these movements or paralysis occurs because of a disability or accident, achieving a proper ankle joint movement is impossible. Owing to the effect of the weakened plantar flexor, inadequate propulsion is produced, which reduces the walking speed [3,4] and shortens the swing phase of the opposite leg and affects the safety of the gait. Meanwhile, damage to the dorsiflexor causes unusual gait patterns [5], [6], [7] because it does not lift the foot at a sufficient angle during the swing phase, and the foot is attracted to the ground, thus making complete walking impossible. These problems affect not only the ankle but also the gait pattern of the entire lower limb, ultimately increasing the metabolic cost [8], [9], [10].

To solve these problems, powered ankle-foot orthosis (PAFO) has been actively developed [11]. PAFO refers to a device that attaches actuators to an orthosis worn on the ankle and can assist the human body. PAFOs include basic science PAFO [12], which studies basic human physiology and biomechanics; augmentation PAFO, which helps healthy people exert more power than usual [13,14]; assistive PAFO, which assists people with disabilities or elderly people who will never walk properly again to walk in a manner similar to that in healthy people [15]; and rehabilitation PAFO, which is used to rehabilitate patients [16]. All these PAFOs are used to control the walking pattern, walking speed, and joint moments by applying various forces to parts of the human body through actuators.

The important factors in the design of these PAFOs, include high force-to-weight ratio, mobility, and alignment with the wearer, referring to the coinciding joints of the wearer and the PAFO [17]. Until now, researchers have focused on the effectiveness of assistance; therefore, they intensively studied PAFOs with high force-to-weight ratios and their effective controls. This is supported by the active use of a series elastic actuator (SEA), which minimizes the weight of the PAFO by connecting the frame to the off-the-board actuator with wire and pneumatic artificial muscle (PAM), which receives air from compressors. Witte et al. developed a SEA with a high bandwidth and applied it to PAFO to implement high-precision torque control [18]. The implementation of this precise control has accelerated the study of assistance effectiveness. It has been actively studied by, for example, finding the optimal point to effectively reduce the electromyography (EMG) or metabolic cost during the gait cycle or heuristic and human-in-the-loop controller of PAFO to determine the optimal assistance profile [18,19]. Because these studies did not seriously consider misalignment, they all used a one-degree-of-freedom (dof) PAFO with the talocrural joint alone as the axis of rotation.

However, the movement of the ankle joint is produced by the rotation of not only the talocrural joint but also the subtalar joint. The ankle joint is a 2-dof motion, which appears to be a 3-dof motion in three dimensions but is realized by a complex rotation of the talocrural and subtalar joints in a twisted position. The talocrural joint plays a major role in plantar flexion and dorsiflexion, producing forward propulsion. The subtalar joint helps to achieve balanced walking on the ground through inversion and eversion. In rehabilitation, simple walking with the talocrural joint is important, although balance training is also essential to strengthen the inversion and eversion movements of the ankle so that patients do not fall by losing their stability [21]. Thus, PAFO also requires the function of the subtalar joint to be used in rehabilitation. The fundamental purpose of using PAFO in rehabilitation is to reduce the fall risk. A fall, which accounts for a large percentage of elderly deaths, is caused by loss of body balance in dynamic or static motion as a result of joints and muscles being weakened by aging, as previously mentioned [22]. Deterioration of balance ability by aging is most related to the weakening of the ankle plantar flexor and decline in ankle eversion range of motion (RoM) [23]. As the plantar flexor, which has the dominant role of supporting the body weight, weakens the forward plantar flexion, the walking speed decreases, and the sway of the center of the pressure and overall stability index increase. The other factor, eversion RoM, also has high correlations with postural sway and fall risk [23,24]. Eversion maintains the center of pressure (CoP) inside the supporting foot, preventing the body from tilting to one side and controlling the direction of the propulsive force of the plantar flexor on unbalanced ground. Therefore, if the RoM of the eversion is not properly secured, the direction of the planar flexor force collapses, and proper balancing cannot be realized. However, the PAFOs that have been developed thus far are 1-dof robots that have one rotation axis alone corresponding to the talocrural joint [11]. It can be useful for weakened plantar flexor but not for reduced eversion RoM. Thus, improvement was needed on the design of PAFO.

The solution for designing PAFO to make up for this insufficient performance is in the detailed explanation of the effects of eversion RoM on balance [25]. Hoogvliet et al. developed a multi-link inverse pendulum model with a rocker-shaped interface and subtalar joint and demonstrated the importance of the subtalar joint on balance control through human experiments. According to the above-mentioned group, that the subtalar joint rotates the body about the frontal plane to compensate the tilting moment resulting from the misalignment between the projection of the center of gravity and CoP when the body is tilted. This action is called the foot tilt strategy (FTS), which can generate a stabilizing moment and return the unstable body to the posture of equilibrium. This also affects the position of the CoP used in the calculation of indexes that indicate balance, such as the extrapolated center of mass and margin of stability [26,27]. CoP can be controlled in the mediolateral direction through FTSs, such as inversion and eversion. Patients who cannot implement frontal plane rotation by arthroplasty of the subtalar joint demonstrate the importance of the FTS in balance control. Therefore, for PAFO to be useful rehabilitation, it should be able to create the stabilizing moment to keep the wearer in balance. For that reason, PAFOs with 1 dof have a contradiction in their fundamental function because they increase the fall risk of elderly wearers who use them as a rehabilitation or assistive device by restricting the subtalar joint. In addition, healthy people who use them as augmentation PAFOs can experience falls or accidents, which means that not all necessary conditions for designing PAFO were met.

A PAFO developed by Roy et al. was used to train and strengthen dorsiflexors in stroke patients [28]. The robot has two hinge joints perpendicular to each other and two linear actuators, allowing rotation to the talocrural and subtalar joints through control. However, this is not a mechanical joint but rather a virtual joint implemented through the control of the actuator. Therefore, control is necessary to limit impossible motion, and the possibility of misalignment increased. To accomplish this, Park et al. fabricated a frame using soft material with no mechanical joint; thus, the possibility of misalignment was close to zero [29]. However, this also imposes a virtual joint by the control of the actuator, and because of the limitation of the soft material, additional limitations arise, such as the difficulty of calculating the kinematics and limited magnitude of the force transmitted from the actuator. These problems can be solved by anatomical data and assigning the actual joint as a mechanical element to the PAFO [30]. Agrawal et al. developed a 2-dof orthosis by calculating the formulas for the talocrural and subtalar joints on the basis of anatomical data [31]. However, their model proved to be simply an unpowered orthosis without actuators to measure the moment on the joint, and in this case, the implementation of the FTS is impossible. We perceived that the FTS function for balance control is an essential element for PAFO and the ability to increase assistance effectiveness. The ability of the FTS enables PAFO to be used for the elderly to rehabilitate or regain their balance skills by providing a certain combination of stabilizing moment and plantar flexion assistance.

To solve the problems of the existing PAFOs, we planned to develop a new PAFO and set the design requirements as follows: The first requirement is i) the implementation of the FTS for enhancing eversion motion with the subtalar joint. This helps potential wearers, who want to be free from fall risk, train the FTS by the rehabilitation of the strength and proprioception of their evertor and balance skills in the frontal plane with the newly developed PAFO, which can assist their eversion. This could not be achieved with the existing PAFOs, which have one joint alone, the talocrural joint. The second requirement is ii) securing assistance effectiveness as much as the degree of the existing PAFOs. As mentioned earlier, the strength of the plantar flexor is also important for balanced walking, so the same function can be conducted with the new PAFO. In accordance with these requirements, we developed a 2-dof PAFO with two rotational axes as mechanical joints by calculating the spatial formulas of the talocrural and subtalar joints by using anatomical data and including them in three-dimensional (3-D) modeling. In addition, two PAMs corresponding to the soleus and peroneus longus were attached to assist the plantar flexor and control the inversion and eversion by controlling the direction of the net force applied to the center of mass of the foot through the control of the magnitude of the force, which makes the FTS possible. We validated, both computationally and experimentally, the design and performance of the fabricated PAFO to determine whether it can be used for its purpose. After validation of the PAFOs, we performed the clinical experiments with healthy participants to verify that the requirements were achieved well with human subjects.

The remainder of this paper is organized as follows: Section 2 describes the design and fabrication of the proposed PAFO. Section 3 discusses the validation of the joint kinematics, workspace, and performance of the actuators of the system. Section 4 shows the mechanical and control characterization of the system, and section 5 addresses the clinical validation of the PAFO in healthy human subjects that shows that the elementary functions for design requirements were properly achieved. Finally, we discuss the potential applications of the proposed PAFO and future studies.