Keywords

1 Introduction

Exoskeletons can be defined as external actuation systems whose purpose is to assist the musculoskeletal system [1]. Although most of these devices focus on user mobility, with complex, bulky, and heavy structures, the trend is for the design of assistive walking devices to become increasingly lighter known as exosuits. These devices are a type of exoskeleton that, using lighter weight actuators, can assist movement, increasing comfort and reducing production costs by eliminating the rigid bars of traditional exoskeletons [1].

Most of these devices are designed primarily to reduce the metabolic cost that the user must perform to carry out an activity, such as walking. The difficulty in the design of these devices depends mainly on the estimation of the assisted joint moment and the way in which the forces are transmitted [2].

To know the impact of the exosuit on the musculoskeletal system, the simulations allow us to calculate muscle activations. In a recent study [3], static optimization was used to calculate muscle activations and the interaction of forces between contact points. Although this is an efficient method of optimization, the absence of correction terms can lead to errors [4]. More recent models propose to relate muscle activations to changes in the model (e.g., position or velocity) so that when considering muscle physiology, the results are more accurate [5, 6]. These simulations are non-invasive tools to quantify the action of the device on the human body [7]. In this sense, the main objective of this work is to evaluate how the use of an exosuit affects the neuromuscular system by studying the muscular forces of the main muscles involved in human gait by varying the anchor point of the device.

2 Materials and Methods

Briefly, we performed a simulation in Opensim [8] to analyze the evolution of muscle activations and the metabolic cost of the cable-driven actuation (Fig. 1).

Fig. 1.
figure 1

Scheme of the simulation for the calculation of activations and muscular efforts

Full size image

2.1 Experimental Data

For the dynamic simulation, we used the data of one participant extracted from a public database by Fukichi et al. [9]. Table 1 shows the data of the subject used for the study.

Table 1. Demographic data of the subject used to run the simulations. Further details in [9, 10]
Full size table

2.2 Dynamic Simulation

Using the kinematic and kinetic data extracted from the previous database [9], the simulations were generated in OpenSim [8] following the scheme shown in (Fig. 2). The first step is defining a generalized model (12 lower limb bodies, and 2 for the exosuit anchor points, each body is attached by a custom joint [11] giving it 20 degrees of freedom. A total of 18 muscle–tendon actuators [5] were able to generate the forces, and 2 additional actuators (PathActuator class) were implemented for the cable actuation controlled by ControlLinear Class [12]). Next, ScaleTool adapt the general model to the anthropometric experimental data. The following step is to calculate the joint angles trajectories using the Inverse Kinematic (IK) tool. To reduce the inconsistency between the ground reaction forces (GRF) and the measured moments with the model kinematics, we apply the Residual Reduction Algorithm (RRA) [8, 13]. Then, the muscle activations were estimated through Computed Muscle Control (CMC) [7, 14].

Cable Force

PathActuator class were implemented [15]. The cable was attached at 0.32 m from the center of the hip joint. The distal part of the cable has been placed at 30%, 50%, and 70% of the total length of the thigh. The desired hip flexion/extension joint moment is known from the inverse dynamics (ID), and the cable force f is:

$$T_{{m \cdot z}} = \left\{ {\begin{array}{*{20}l} { - r_{{AP}} \times f|_{z} } \hfill & {{\text{if}}\;\tau _{{m,z}} > 0} \hfill \\ {0,} \hfill & {{\text{if}}\;\tau _{{m,z}} \le 0} \hfill \\ \end{array} } \right.$$
(1)

where τm.z is the z-component of the joint moment, rAP is the vector of the cable anchored to the thigh. The value 0 means that the cable is in compression.

3 Results

Through the developed simulation framework, the data on muscle activations and metabolic costs are obtained. The results of these are discussed in the following sections.

3.1 Muscle Activations

The muscle activations decrease as the position of the anchor point is further away from the center of the hip joint. As can be seen in Fig. 2, in the hip muscles there is a reduction in muscle activation levels as the exosuit anchor point is positioned further away from the proximal part of the thigh (e.g. vastus lateralis and semitendinosus). This decrease in muscle activations occurs when exosuit begins to act. During the phases prior to exosuit action, muscle activations remain unchanged.

Regarding the muscles of the lower leg (e.g. tibialis anterior and soleus in Fig. 2), the exosuit would not modify the dynamics of the lower leg, and therefore, the kinematics imposed by the subject would not be affected in any way.

Fig. 2.
figure 2

Muscle activations of the actuating muscles of the model for each of the positions

Full size image

3.2 Metabolic Cost

As the performance is further away from the proximal hip, a reduction in metabolic cost is seen Fig. 3. It should be noted that, although at the 30% level, there is a reduction in muscle activation, the total metabolic cost is increased compared to the non-actuated gait (1.65%). When the exosuit is placed at mid-thigh, i.e., 50% position, the total metabolic cost is reduced by 6.36% and at 70% position, it is reduced by 10.73% with respect to the metabolic cost of a non-actuated gait.

4 Discussion

In this work, dynamic simulations were generated to obtain the influence of the anchor point, and compare its influence on the muscle activations and metabolic cost. Although this model only includes movement in the sagittal plane, similar results have been obtained with respect to the previous work of Dembia et al. [16]. These authors performed a simulation with a more complex model during an ideal actuator-assisted gait cycle and observed a greater reduction in muscle activation in those muscles that act in more than one degree of freedom (biarticular muscles). In our study, it is shown that muscles such as the semitendinosus muscle show a lower level of activation when compared to the reference simulation in which it is not actuated. For the muscles that are not being actuated, there is no change in the actuation dynamics, therefore, the actuation of the exosuit does not influence those actuators that are not involved in the movement of the hip joint 2. An example of the use of the exosuit for rehabilitation in older adults is the one developed by Jin et al. [17]. In this study, the authors conducted a trial with an older adult walking on a small slope with a cable actuated exosuit, obtaining a significant reduction in metabolic consumption of 7.7%, value similar to our results (Fig. 3). The results of our simulations show that the performance of the exosuit reduces the muscle actuation required by the actuator muscles in the assisted joint, without influencing the dynamics of the others, aiding that would not influence the gait pattern of the subjects.

Fig. 3
figure 3

Average reduction in metabolic cost at different levels of action represented by the red border. In blue, represents the variation in the percentage of metabolic cost

Full size image

5 Conclusion

A model has been used for the simulation of performance through an exoskeleton (exosuit) in which the activations of 9 muscles are evaluated during a gait cycle. In this work, we propose a study of the influence of the placement of the exosuit performance cable on the person’s thigh by studying its influence in various positions.

Our results suggest that as the cable is placed more distal to the hip joint, the activation level exerted by the hip flexor and extensor muscles decreases in the exosuit actuation phase. These simulations serve as a basis for the construction of a wearable exoskeleton to improve the capabilities of the elderly.