Walking with a powered ankle-foot orthosis: the effects of actuation timing and stiffness level on healthy users

Background In the last decades, several powered ankle-foot orthoses have been developed to assist the ankle joint of their users during walking. Recent studies have shown that the effects of the assistance provided by powered ankle-foot orthoses depend on the assistive profile. In compliant actuators, the stiffness level influences the actuator’s performance. However, the effects of this parameter on the users has not been yet evaluated. The goal of this study is to assess the effects of the assistance provided by a variable stiffness ankle actuator on healthy young users. More specifically, the effect of different onset times of the push-off torque and different actuator’s stiffness levels has been investigated. Methods Eight healthy subjects walked with a unilateral powered ankle-foot orthosis in several assisted walking trials. The powered orthosis was actuated in the sagittal plane by a variable stiffness actuator. During the assisted walking trials, three different onset times of the push-off assistance and three different actuator’s stiffness levels were used. The metabolic cost of walking, lower limb muscles activation, joint kinematics, and gait parameters measured during different assisted walking trials were compared to the ones measured during normal walking and walking with the powered orthosis not providing assistance. Results This study found trends for more compliant settings of the ankle actuator resulting in bigger reductions of the metabolic cost of walking and soleus muscle activation in the stance phase during assisted walking as compared to the unassisted walking trial. In addition to this, the study found that, among the tested onset times, the earlier ones showed a trend for bigger reductions of the activation of the soleus muscle during stance, while the later ones led to a bigger reduction in the metabolic cost of walking in the assisted walking trials as compared to the unassisted condition. Conclusions This study presents a first attempt to show that, together with the assistive torque profile, also the stiffness level of a compliant ankle actuator can influence the assistive performance of a powered ankle-foot orthosis.

: Schematic drawing of the MACCEPA and its parameters [1]. B and C are the distances from the joint axis of the attachment points of the MACCEPA spring on the leverarm and the output link, respectively; k and prec are the MACCEPA spring constant and pre-compression, respectively; α is called the deflection angle; ϕ is the equilibrium position angle, also called leverarm angle. A(α) is calculated as shown in Eq. 2.
The MACCEPA is a torque-controlled, variable stiffness actuator (VSA) whose working principle is described in details in [1]. The schematic drawing of the MACCEPA and the MACCEPA's parameters are shown in Fig. 1. As shown in Fig. 1, the MACCEPA consists of three bodies (fixed link, output link, and leverarm) pivoting around a common axis. A spring is attached between the leverarm and the output link. The torque (τ ) provided by the MACCEPA is defined by its parameters as shown in Eq. 1: where A(α) is calculated as follows: In the MACCEPA, the configuration in which the leverarm and the output link of the actuator are aligned, thus, in which the deflection angle (α) is equal to zero, is called the equilibrium position. In this configuration, the force created in the MACCEPA spring by the spring pre-compression (prec) generates no torque in the actuator, due to the alignment of the leverarm and output link. Figure 2 shows the MACCEPA-based actuator used in the experiments presented in the article. More details on the actuator's design can be found in [2].
The spring pre-compression (prec) in Eq. 1 is measured in meters. However, the level of spring pre-compression (P ) is defined as a percentage of the working length of the MACCEPA spring. Specifically, the level of spring pre-compression is defined to be equal to 0% when the length of the spring at the actuator's equilibrium position is equal to the length of the uncompressed spring (i.e. the natural length of the spring). Similarly, the level of spring pre-compression (P ) is equal to 100% when the length of the spring at the equilibrium position is equal to the length of the spring when it is completely compressed.
The pre-compression of the MACCEPA spring modifies the behavior of the MACCEPA. Higher spring pre-compression levels result in stiffer behavior of the actuator. On the contrary, the actuator is more compliant with lower spring pre-compression levels. In other words, the MACCEPA spring pre-compression determines the amount of torque exerted by the MACCEPA (with fixed values for the parameters B, C and k) for a fixed value of the deflection angle. Figure 3 shows the effect of the spring pre-compression level (P ) on the behavior of the MACCEPA-based actuator (Fig. 2), i.e. its deflection angle-torque characteristics.
As it can be seen from Fig. 3, for each pre-compression level, both the MACCEPA output torque (τ ) and apparent stiffness are non-linear functions of the deflection angle (α) [3]. The deflection angle-torque characteristics can be approximated with a third order polynomial, thus, the deflection angle-stiffness characteristics can be approximated using a second order polynomial as presented in [3]. The resulting approximated deflection angle-stiffness characteristics of the MACCEPA-based ankle actuator for the spring pre-compression levels used during the experiments (20%, 40%, and 60%) are shown in Fig. 4