Assist region
To determine the appropriate target assist region, biological hip moment data were analyzed. Running motion data from seven participants running at 2.5 m/s, provided by Kim et al. as supplementary material [4], were used for the analysis. The average hip moments according to the hip angles, along with the average hip moment and angle during the running gait, are shown in Fig. 1. Based on the biological hip moment data, the hip joint is required to generate flexion moments in the range of motion between a flexion angle of 18° and maximum extension; this region is marked as yellow in Fig. 1. Moreover, the flexion moment in this region has a moderately linear correlation with the extension of the hip angle (R2 = 0.53 for the region with the hip flexion moment).
Considering this biological feature, the exosuit was designed to passively exert flexion moments on the hip joint by using the elastic bands placed in front of each thigh while the hip joint moved in this range of motion. The elastic band was designed to be in a slack state at hip angles between a flexion angle of 18° and maximum flexion to ensure that the band did not exert any moment on the hip. When the hip is inclined at an angle of 18° in the flexion direction, the band starts to elongate and generates a passive assistive moment on the hip joint. Moreover, the passively exerted moment was designed to increase as the thigh moved toward maximum extension with the increase in the resilience of the elastic band due to its elongation.
Suit design
An exosuit capable of assisting hip flexion was fabricated using textiles and elastic bands, as shown in Fig. 2. To minimize the suit deformation from the force of the band, a composite fabric-based textile (CT5K.18/wov4, Dyneema, USA) was used as the base material owing to its high toughness and low weight. Another layer of thin polyester-nylon-based textile was placed on the sides facing the body to reduce slippage between the exosuit and clothes worn under it. Tight-fitting sweatpants made of polyester were required to be worn underneath the thigh pieces to enhance the tactile sensation and prevent slippage. The “Exoband” fabricated by Panizzolo et al., which could provide hip flexion moments for the elderly and facilitate walking, was referenced in the design of the proposed suit [12].
The exosuit consisted of five parts: two thigh pieces, one torso harness with shoulder straps, and two band modules that could be attached between the torso harness and each thigh piece in the upper and lower band grounding positions, as depicted in Fig. 2A. Because the body sizes of users may vary considerably, the length of the band modules was designed to be adjustable using a ratchet system to enable customization. Furthermore, the shoulder straps and radius of the torso harness and the radii of the thigh pieces were designed to be adjustable to ensure that the suit can fit different body shapes. The loadcell was attached to the band module to measure the forces from the band exerted on the exosuit during running. The fabricated exosuit weighed 609 g, and the specific weights of the device components are presented in the supplementary material.
Suit deformation analysis
Owing to the forces exerted on the suit by the band, the suit undergoes deformation that is nonlinear compared to the exerted force. This deformation may result in less elongation of the band than intended if not considered in advance. To estimate the effects of the band on the hip joint in the design process, this deformation must be considered. Therefore, a simulator that could estimate the assistance profile of a band according to the hip angle was developed.
Kinematic modeling of the human thigh with the flexion assistance force in the sagittal plane of the leg (Fig. 3), as suggested by Wei et al. [13], was performed to calculate the displacement between the upper and lower grounding positions on the body according to the hip angles. Parameters a1 and a2 were set as 0.15 m and 0.3 m, respectively, and c1 and c2 were set as 0.12 m and 0.08 m, respectively, based on the size of the human body and upper and lower band grounding positions on the body. Through a1, a2, c1, c2, and \(\theta\), the initial length of the band module, \({l}_{0}\), and length of the band module at hip angle \(\theta\), \({l}_{\theta }\), could be calculated using trigonometric principles. The displacement \({\delta }_{total}(\theta )\) was expressed as in Eq. (1).
$$\delta_{total} \left( \theta \right) = l_{\theta } - l_{0}$$
(1)
For the proposed exosuit, \({\delta }_{total}\) involved two parts, as expressed in Eq. (2), namely, elongation of the band and suit deformation.
$$\delta_{suit} + e_{band} = \delta_{total}$$
(2)
where \({{\varvec{\delta}}}_{{\varvec{s}}{\varvec{u}}{\varvec{i}}{\varvec{t}}}\) is the displacement from the deformation of the suit and \({{\varvec{e}}}_{{\varvec{b}}{\varvec{a}}{\varvec{n}}{\varvec{d}}}\) is the elongation of the elastic band. This relationship was required in the design process of the band modules to provide the appropriate assistance. The following section presents the derivation of the two terms.
Suit deformation model
To compensate for the suit deformation, we formulated the suit stiffness model by using the measurement methods suggested by Lee et al. [14]. Specifically, we used a BLDC motor (EC-4 pole 30-305015, Maxon Motor, Switzerland) with a gearhead (GP 32 HP-326664, Maxon Motor) to pull a Bowden cable (FCP-04DB, RESPONSE, Taiwan, used with cable housing BHL100, Jagwire, Taiwan) connecting the upper band grounding position of the torso harness and lower band grounding position of the thigh pieces receiving the force from the band module, as shown in Fig. 4. The received force was directly measured from the cable by using the loadcell. The deformation of the suit was estimated to be the displacement of the cable position from its initial position, caused by pulling from the motor. The cable position at which the loadcell data changed from zero to a positive value was set as the initial position of the suit before deformation. In the test, it was assumed that the participant was completely still and that no extension occurred in the length of the Bowden cable.
To acquire the data for deriving the suit stiffness model, we tested the device on three male participants (age: 24.8 ± 0.9 y; height: 1.72 ± 0.08 m; weight: 75.3 ± 6.8 kg) while they stood in a posture with a hip extension of 12°, which is similar to a 48% gait cycle (GC) in running. We selected this posture because the exosuit was designed to exert maximum force on the wearer at this point of the GC. The received force and deformation of the suit were recorded from each participant while the cable was pulled at a constant rate of 0.05 m/s until the peak force reached 100 N. We repeated the tests 10 times and used all the recorded data to derive the stiffness model.
The suit stiffness model was derived by fitting the recorded data to Eq. (3) proposed by Lee et al. [14].
$$\delta_{suit} = C_{s1} \ln (C_{s2} F + 1)$$
(3)
where \({\delta }_{suit}\) is the deformation of the suit and Cs1 and Cs2 are the model coefficients. The original test data and fitted curve using Eq. (3) are shown in Fig. 4.
Band selection for three-level assistance
According to previous studies, the optimal assistance level varies among individual participants when using supportive devices [15,16,17]. Therefore, we tested three bands with different levels of stiffness to exert different levels of assistive force on the participants. The design parameters of these bands were selected considering the suit deformation.
The estimated decrease in the maximum flexion moment during running at 2.5 m/s was used as the index for the assistance levels of the band modules. The participants tested band modules with different stiffness values, and a band module estimated to reduce the maximum flexion moment by 6.5% was set to correspond to the highest stiffness. This selection was based on feedback from a pilot study that indicated that bands stronger than this threshold induced discomfort during running. In addition to this band, bands expected to reduce the maximum flexion moment by 5.5% and 4.5% were selected as the test bands.