Configuration Design and Analysis of Reconfigurable Supernumerary Robotic Legs for Multitask Adaptation: SuperLegs

Supernumerary robotic limbs (SRLs) are a new type of wearable robot that add artificial limbs to the human body to perform collaborative tasks. In contrast with exoskeletons, SRLs are kinematically independent of human limbs, allowing the wearer to overcome the limitations of human physiological ability, such as realizing the expansion of space in the human body, rather than enhancing existing limbs. In this study, a lightweight and compact hexagonal reconfigurable lower‐limb SRL system is proposed to assist human locomotion in daily activities, including walking, crouching, and stair climbing. To adapt to multiple scenarios, the hexagonal mechanism can be adjusted to different configurations including convex hexagonal configuration, pentagonal configuration, and concave hexagonal configuration. To achieve more optimized performance of different configurations, the optimization for identifying the optimal dimensions of each link was carried out. Subsequently, the detailed design methodology and specifics are presented. Finally, the load and wearing performance experiments were evaluated. The experiment results demonstrated that the tested maximum load in different configurations exceeded 90% of the simulated value and the entire equipment has a good wearing adaptability. This study may inspire the design of other lower‐limb SRLs and provide efficient solutions for stable support assistance in various scenarios.


Introduction
Wearable robots have been extensively and comprehensively studied in military, civil, and medical-related fields among other areas of study.There are two categories of wearable robots: exoskeletons and robotic prostheses. [1,2][5] The development of exoskeletons and prostheses has been studied comprehensively over the last decades, before the concept of supernumerary robotic limbs (SRLs), a novel type of wearable robot, was proposed.SRLs were first introduced in 2012 by Davenport et al., who proposed two additional arms to perform tasks together with the wearer's arms. [6]esigned to achieve human sensorimotor augmentation by enhancing physical capabilities for manipulation and locomotion tasks, SRLs have several unique advantages in terms of technical advancement and performance.Artificial degrees of freedom (DOFs) are added to the human body to facilitate the performance of relatively complicated activities with enhanced strength and precision. [7]10] SRLs are classified into three branches: supernumerary robotic arms, supernumerary robotic legs (SRLGs), and supernumerary robotic fingers, each of which is equipped with functionalities similar to those of corresponding human limbs. [11]upernumerary robotic limbs (SRLs) are a new type of wearable robot that add artificial limbs to the human body to perform collaborative tasks.In contrast with exoskeletons, SRLs are kinematically independent of human limbs, allowing the wearer to overcome the limitations of human physiological ability, such as realizing the expansion of space in the human body, rather than enhancing existing limbs.In this study, a lightweight and compact hexagonal reconfigurable lower-limb SRL system is proposed to assist human locomotion in daily activities, including walking, crouching, and stair climbing.To adapt to multiple scenarios, the hexagonal mechanism can be adjusted to different configurations including convex hexagonal configuration, pentagonal configuration, and concave hexagonal configuration.To achieve more optimized performance of different configurations, the optimization for identifying the optimal dimensions of each link was carried out.Subsequently, the detailed design methodology and specifics are presented.Finally, the load and wearing performance experiments were evaluated.The experiment results demonstrated that the tested maximum load in different configurations exceeded 90% of the simulated value and the entire equipment has a good wearing adaptability.This study may inspire the design of other lower-limb SRLs and provide efficient solutions for stable support assistance in various scenarios.
Compared with traditional lower limb exoskeletons, SRLGs can prevent the interference of legged exoskeletons and provide extra support without constraining the legs because of the kinematic independence of human limbs.][15] For loaded walking, Hao et al. developed a robotic system, SuperLimb, that transferred the payload to the ground through two robotic limbs attached to the user's back. [12]Parietti et al. proposed SRLGs constituting two straight robotic legs with three DOFs to provide balance assistance and reduce joint load for human bipedal walking. [13]Khazoom et al.
proposed SRLGs designed for assisting various gaits during walking, where delocalized magnetorheological clutches were used to reduce inertia during swinging. [14]All proposed SRLGs are designed in series configuration and a parallel reconfigurable configuration scheme has not been adopted for the SRLGs design.[18] Zhang et al. developed a series-parallel-reconfigurable tendon-driven upper limb SRLs for various tasks, such as performing overhead tasks, payload carrying, and collaborative work with human arms. [16]Zhang et al. introduced an upper limb exoskeleton with a five-bar linkage for lifting assistance and loading capacity and the maximum payload reached 1.5 times its weight. [17]Parallel mechanisms have been extensively studied in the last decades, and their advantages over serial links have been demonstrated.Parallel robots usually have higher stiffness and movement accuracy because the mass carried by each link and the errors in the actuator positions are averaged. [19]Furthermore, the load-weight ratio is significantly increased. [20]urrent SRLG prototypes are primarily designed to assist the operator in a single scenario, in which case their mechanical design and control algorithms are specifically developed.There is a demand for versatile SRLGs suitable for various tasks.Parallel configurations are potentially beneficial in terms of adaptation to multiple scenarios because the geometric shape of the parallel structure can be reconfigured, the potential benefits of which are still beyond exploration. [21,22]The reconfigurable mobile robot developed by Li et al. is equipped with three locomotion modes and seven gaits based on motion branch variations, making it relatively practical in various scenarios. [21]he two-DOF parallel robot designed by Rahul et al. formed a planar pentagon with five bar linkages and the gripper connected to the robot covered a wide range by reconfiguring the shape of the pentagon. [22]Considering the features of parallel configurations, a lightweight and compact hexagonal reconfigurable lower limb SRL system (SuperLegs) is proposed.The system can be transformed into different configurations including convex hexagonal configuration, pentagonal configuration, and concave hexagonal configuration for multitask adaptation.In this article, we present configuration design and analysis of SuperLegs which could support several activities of daily living (ADLs), including walking, crouching, and stair climbing.At the same time, the optimization of link length is carried out to improve the load output in the optimized workspace for various ADLs and the detailed design methodology and specifics are presented.The diagram of SuperLegs is illustrated in Figure 1.The main contributions of the work presented in this article are summarized as follows: at present, almost all proposed SRLGs were designed in series configuration and used in a single scene.No researcher adopted a reconfigurable configuration scheme for SRLGs design.The proposed SuperLegs can be adjusted to different configurations including convex hexagonal configuration, pentagonal configuration, and concave hexagonal configuration.The reconfigurable configuration design enables SuperLegs to better adapt to multiple scenarios, such as walking, crouching, and climbing stairs.The proposed configurations are well implemented.The payload evaluation experiment results demonstrated that the SuperLegs has an effective load assistance ability under variable tasks.This study may inspire the other SRLGs and provide efficient solutions for stable support assistance in multitask adaptation.
The remainder of this article is organized as follows.In Section 2, the configuration design is introduced.In Section 3, the characteristic analysis of SuperLegs is discussed.In Section 4, the mechanical design of SuperLegs is discussed.In Section 5, the implementation and evaluation of SuperLegs are discussed.Finally, conclusions and outlook are presented in Section 6.

Configuration Design
A hexagonal reconfigurable configuration scheme is proposed.The SuperLegs has three configurations including convex hexagonal configuration, pentagonal configuration, and concave hexagonal configuration that correspond to the four ADLs, as shown in Figure 2. When walking on flat ground, both joints of the foot pedal are in contact with the ground, and the rollers mounted on joints 1 and 2 roll on the ground.When walking on rough ground, the convex hexagonal configuration alternately cooperates with the human feet to better adapt to the rough ground environment.The other two ADLs were stair climbing and crouching.Given the limited stair dimensions, the convex hexagonal configuration was degraded into a polygonal configuration for the stair climbing scenario, supporting the wearer with only one joint in contact with the ground.As shown in Figure 2d, the lower part of SuperLegs folds into a concave hexagonal configuration.Two thigh links support the weight and can adjust the height by changing the distance between joints 2 and 5 until they form a trapezoid.

Analysis of the Configuration Adaptation for Multiple Tasks
Multiple configurations are designed for multiple scenarios and each configuration has a unique adaptable scenario.Conversely, in a certain scenario, other configurations cannot replace the corresponding configuration.For the walking task, first, the convex hexagonal configuration has high stability and a large area of the support polygon compared with the pentagonal configuration.Second, the convex hexagonal configuration can resist torque from the load.Finally, when walking on flat ground, the convex hexagonal configuration allows the two-joint wheels to come into contact with the ground and provide a relatively stable march.For stability, we assumed that the SuperLegs foot pedal and human foot are parallel to each other and that the footplate is simplified by line segments.As shown in Figure 3, the areas of the stability domains for the convex hexagonal and pentagonal configurations are expressed as follows where S 1 and S 2 represent the support polygon area of the convex hexagonal and pentagonal configurations, respectively; d 1 represents the vertical distance between the center of the adult left and right foot; d 2 represents the vertical distance between the SuperLegs foot pedal and the ipsilateral human foot; and L foot represents the length of the foot.The stability condition for the two configurations is that ZMP point always exists inside of the support polygon. [23]P where ZMP 1 and ZMP 2 represent the ZMP points of the convex hexagonal and pentagonal configurations, respectively, and can be approximately estimated through the motion state. [23]Using Equation (1), we can obtain Therefore, the convex hexagonal configuration has relatively high stability.For stair climbing, the convex hexagonal configuration can easily interfere with the stairs, thereby seriously affecting the flexibility of assistance by the system and comfort of the wearer.Therefore, the pentagonal configuration exhibited more optimized flexibility and auxiliary efficiency.For the crouching task, compared with the other two configurations, first, the concave hexagonal configuration acts as a four-point support system, and the four legs can move alternately, yielding higher stability than that of other configurations.Second, the concave hexagonal configuration had a higher output capacity in the crouching space.A simulation comparison experiment, where the weights of the structure and the motors were disregarded, was conducted.The link lengths were 320, 600, 500, 302, 500, and 600 mm.The crouching auxiliary space is defined as a concave hexagonal reachable space.The force analysis results for different configurations are shown in Figure 4.The richness of color represents the maximum static output force of each configuration.For the concave hexagonal configuration, as shown in Figure 4a, the maximum force in most areas exceeded 180 N.For the convex hexagonal and pentagonal configurations, as shown in Figure 4b,c, the maximum force in most areas exceeded 125 and 90 N, respectively.Compared with Figure 4a-c, notably, the concave hexagonal configuration is more advantageous in terms of the output force for the crouching scenario.

Kinematic Analysis
The detailed modeling of SuperLegs in different configurations is depicted in Figure 5.The green markings in the schematic are the actuated joints and the rest are passive joints.the entire system in different configurations is represented by J walking , J climbing , and J crouching .

Convex Hexagonal Configuration
A kinematic description of the convex hexagonal configuration is presented in Figure 5a.To solve the forward kinematics of the convex hexagonal configuration, the positions of P 1 and P 2 , which determine the position and direction of link 4, need to be determined.The position expressions for P i are presented in Appendix A. The end position of P can be obtained by P 1 and P 2 as follows The end orientation of link 4 can be computed as follows Then, the passive angle θ 4 , θ 2 in the above equation should be represented by the active angles θ 1 , θ 3 , θ 5 and the expressions of θ 4 , θ 2 are presented in Appendix A.
The inverse kinematics of convex hexagonal configuration consists of the following two steps: 1) derivation of P 1 and P 2 from the given position and orientation of end link 4; 2) resolution of inverse kinematics of serial chain.

Pentagonal Configuration
A kinematic description of the pentagonal configuration is shown in Figure 5b.The positive position analysis equation can be obtained using the following constraint equation The passive angles θ 2 , θ 4 in the above equation are represented by active angles θ 1 , θ 3 .The expressions for θ 2 , θ 4 are presented in Appendix B.
Knowing the end position p ¼ p x , p y , according to the knowledge of plane geometry and the law of cosines, the formulas of θ 1 and θ 4 are calculated as follows where θ 1 and θ 4 represent the active joint angles and the specific expressions for α 1 , α 2 , and α 3 are presented in Appendix B.

Concave Hexagonal Configuration
A kinematic description of the pentagonal configuration is shown in Figure 5c.The positive position analysis equation can be obtained using the following constraint equation The specific expressions of β 1 , β 2 , and δ 2 are presented in Appendix C.
In this section, the inverse solution is obtained using link 5, which remains parallel to the ground.First, given the end position and attitude, the positions of P 1 and P 2 can be expressed as follows 8 < : Subsequently, based on the knowledge of plane geometry and the law of cosines, the formulas for the active joint angles θ 1 , θ 3 , and θ 5 were calculated as follows 8 > < > : The specific expressions of α 1 , α 2 , α 3 , and β 1 are reported in Appendix C.

Optimization of Links Length
To improve its performance in terms of a suitable workspace and large load output for various ADLs, optimization for identifying the optimal dimensions of each link should be performed so that the system can provide satisfactory multitask adaptation.To guarantee its performance and usability, the constraints for the topological design are summarized as follows: 1) The foot pedal movement should cover the space of the wearer's foot position during various ADLs.2) To achieve the crouching configuration, the lengths of links 2 and 6 should exceed those of links 3 and 5.
3) The dimensions of each link should not be excessively large because the longer the links, the larger the size of the entire system.4) The dimensions of links 2 and 6 should not be excessively long, which will increase the height of the hip position while crouching, making the adjustable range unable to cover the typical crouching movement.5) For each set of link parameters, SuperLegs should be fully accessible to the optimization space.
[26][27] The optimization results have been significantly improved and the necessity of optimization has been highlighted.For this kind of power-assisted wearable robot proposed in this article, the output capacity within an effective assistance space is an important optimization index.The optimization of link length was carried out for the design of SuperLegs by selecting the average output force in the walking, climbing, and crouching conditions as the objective.By determining the maximum torque of the SuperLegs active joints, the maximum supporting force of any point in the motion space can be obtained using the following formula where F max is the output force; J is the Jacobian matrix relating endpoint velocity to joint velocities, represented by J walking , J climbing , and J couching in the three operating situations.τ max represents the output torques of active joints.Based on ergonomic data in ref. [28], the optimized workspaces of the three motion configurations are selected, as shown in Figure 5. Considering that the main auxiliary output of SuperLegs is in the direction of gravity, the output force in the y direction is selected as the optimization target.The optimization process maximizes the objective function over parameter , where L 1 , L 2 , L 3 , L 4 , L 5 , L 6 are the length parameters of the 6-bar linkage.The specific optimization objective function is expressed as follows where F walking is the average output force in the optimized walking workspace, F climbing is the average output force in the optimized climbing workspace, and F crouching is the average output force in the optimized crouching workspace.w 1 , w 2 , and w 3 are the weight coefficients of the parameters that are empirically set according to the usage rate of the scenario.In this study, we focused on the application of SuperLegs on weight-bearing walking and rarely applied it to crouching.Therefore, the specific results of w 1 , w 2 , and w 3 are 0.6, 0.3, and 0.1, respectively.As too many input parameters significantly increased the response time of the optimization program, the current model was simplified by controlling the number of variables, which captured the influence of each parameter on the outcome separately.
The initial geometric dimensions of each link are presented in Table 1.During the single-variable optimization, the others were assigned a constant so that the effect caused by the specific link length was identified.Within the available parameter range, the objective function ( 14) was maximized by optimization.
Figure 6 shows the relationship between the objective function and length change of the foot pedal L 4 and hip link L 1 , separately.
Although both yielded more optimized results with shorter lengths, this was not feasible because of constraints.
Regarding the minimum necessary length of the hip link, which allows for the installation of three motors, the final length for L 1 was 300 mm.In addition, because the foot pedal should be sufficiently long to fit most adult foot sizes, the length of L 4 after optimization was 300 mm.A symmetric structure, L 2 = L 6 ¼ l thigh and L 5 = L 3 ¼ l shank , was adopted for SuperLegs.The optimization parameter after simplification was P ¼ l thigh , l shank Â Ã .A modified particle swarm optimization (PSO) algorithm with a double-check procedure for interrelated inputs was designed to determine the overall optimal values for the thigh and shank links.A process similar to that outlined in a previous study based on basic PSO that set up time-varying linear inertia weights [29,30] was applied to facilitate the occurrence of global convergence of all particles at the early stage, while the local convergence  capability at the end was reinforced to precisely locate the optimal value.PSO is typically used as an independent variable; however, because the link parameters for SuperLegs are coupled owing to the bonded sum of the thigh and shank links, we added a double-check sequence when updating the values for each particle.During initialization, particles with inappropriate values were regenerated until they met the selection criteria.After calculating the position and velocity of each particle using Equation ( 15) and ( 16), all results were censored and those that exceeded the boundary of the total length were randomly regenerated, thereby slightly increasing the required number of iterations but with a minor negative impact on the final results.
where n is the number of iterations, which is set to 50;

Mechanical Design
The mechanical design of SuperLegs is shown in Figure 9.The system consists of two mechanical limbs located at the waist and a loaded backpack.Through the transformation of configurations to adapt to different tasks, movement can be coordinated with human lower limbs to assist the wearer in walking with heavy loads, ease the burden, and provide support under special unsuitable configurations, such that the human body can maintain stability and reduce the sense of weight-bearing under heavy loads.Each mechanical limb was designed as a six-link closed-chain parallel mechanism with three DOFs.Through cooperative control of the three active joints, the six-link mechanism can facilitate the expected translation of the implementation follower in the xand y-directions on the plane and rotation around the z-axis.In special cases, the joint can be locked and the mechanical limb reduced to a five-link mechanism to achieve point contact with the ground.The DOF of the mechanical limb system was calculated according to the Kutzbach-Grübler formula as follows where m denotes the DOF of the rigid body (for the planar mechanism, m = 3), N denotes the number of components (including the frame), n denotes the number of joints, and f i denotes the DOF of the corresponding joints.A hollow carbon fiber tube, which has the advantage of being lightweight while ensuring stiffness and robustness, was used as the connecting rod.The three driving modules are centrally arranged on the frame of the waist side link, making the mechanical structure compact, while keeping the motor relatively static with the human body, thereby reducing the inertia of the system.To improve the overall coordination and output performance of the system, active and passive joints were arranged in a crossover.The active joint was driven using a remote transmission system.Active joint 1, as shown in Figure 9a, was driven by a synchronous belt wheel designed using an external tensioning device.The synchronous belt wheel at the active end had 24 teeth, whereas that at the driven end had 40 teeth.The synchronous belt wheel at the driven end was fixed and connected to the shaft of the active joint using a key.The transmission ratio was 1.67; therefore, the rated torque designed at active joint 1 was 15 N.m.As shown in Figure 9b, first, a gear set with a transmission ratio of 2 is used to amplify the torque of the active end, and then the output force is transmitted to the far end through a double parallelogram linkage mechanism.The red triangle is an entity whose two sides form one side of two series parallelograms to realize the drive of the remote cross-active joint 2. The rated torque designed at active joint 2 was 18 N.m.As shown in Figure 9c, the drive system amplifies the torque using two sets of gears with a transmission ratio of 4 and the remote force transfer is achieved through the synchronization belt.The rated torque at active joint 3 is 36 N.m.The weight of the overall one-legged SuperLegs system was 7.6 kg and the average output payload of three configurations after optimization was 16.28, 19.54, and 25.14 kg, respectively.Table 2 summarizes the other SRLGs that mentioned weight and payload output.It is obvious that the proposed SuperLegs is relatively lightweight and advantageous in payload capacity.Khazoom et al. [15,31] 1 9.7 5 Yang et al. [32] 2 14.5 15 Hao et al. [12] 2 8.7 7.5 Gonzalez et al. [33,34]

Implementation of SuperLegs
Figure 10 shows the implemented SuperLegs.The transmission ratios of the three active joints were set to 1.67, 2, and 4 actuated by AK80-9 (230 W, 9 N•m, 485 g) mounted near the backside.To minimize the remote mass worn by the user, along with the foot pedal and hip link, other links were fabricated using standard carbon fiber tubes.Most of the other components were made of an aluminum alloy (AL6061-T6).A single side of the SuperLegs weighed 5.2 kg, including the structures, motors, and sensors.SuperLegs is a wearable system with a backplane, controller, and battery, resulting in an additional 2.4 kg. Figure 10b-d shows the wearing effect under the three configurations.It is noticed that the convex hexagonal configuration can adapt well to the change in the walking center of gravity, the pentagonal configuration can reasonably match the geometric dimensions of stairs, and the concave hexagonal configuration can adapt to different crouching states.Therefore, the implemented SuperLegs has the potential ability of practical application for multiple tasks.

Payload Capability Evaluation
To verify the static output capacity of the SuperLegs described in the simulation results above after optimization, payload experiments with different auxiliary configurations were conducted on the SuperLegs without wearing them.Figure 11a shows the experimental setup for different auxiliary configurations.The payload experiments were composed of SuperLegs, an ATI force sensor, a mechanical platform on which the force sensor could be moved horizontally, and a guide wheel to maintain the vertical load direction.Under the position loop, SuperLegs retains the configurations shown in Figure 11b.The feedback torque values of all the driving motors were checked when the force sensor moved horizontally to the left.When the feedback torque of any driving motor reached the rated torque value, the value of the force sensor was recorded as the maximum load.The experiment was repeated 3 times and the average of the three experimental values was regarded as the final experimental result.
Figure 11b shows the tested and simulated maximum load results for the different load test configurations.Compared with the simulated maximum load, the tested maximum load in all configurations exceeded the 90% of the simulated value.The difference between the tested and simulated results may have been caused by disregarding the weight of the structural parts in the calculation.

Wearing Performance Evaluation
To test the adaptability between the device and the wearer, a wearing experiment was conducted.In the experiment, three male subjects of different heights were invited to participate as the operator.
The heights of the subjects are 170, 175, and 181 cm, respectively.The weights of the subjects are 62.5, 67.5, and 65 kg, respectively.Each subject repeated the wearing process 3 times and the wearing time was recorded every time.Figure 12 demonstrates the wearing time of different subjects.The wearing time of different subjects was completed within 25 s, indicating that the entire equipment has good wearing speed and good wearing adaptability to different heights and weights of users.

Conclusions and Outlook
This study presents a hexagonal reconfigurable lower limb SRL system.The hexagonal mechanism can be reconfigured into different auxiliary configurations to adapt to different scenarios such as walking, crouching, and stair climbing.Considering the maximum sum of the output forces in the appropriate workspace under the three configurations as the optimal objective, a modified PSO algorithm was used for optimization to obtain the optimal link length.Subsequently, a detailed design of the ontology structure was developed.Finally, the load capability experiment was conducted to verify the static output capacity of SuperLegs after optimization.The wearing experiment was conducted to test the adaptability between the device and the wearer.
The experimental results demonstrated that the tested maximum load exceeded the 90% of the simulated load under different configurations and the equipment has good wearing adaptability to different heights and weights of users, indicating great potential for future applications.At present, because our research is still in preliminary stages, real human-wearing-assisted experiments have not been carried out yet.In the next work, we will carry out detailed modeling and identification of the system dynamics and use an end-force sensor to realize the precise control of the foot force.Then, the relevant sensing suite will be used to obtain the gait of the wearer so that SuperLegs can make reasonable adaptations and conduct research on human-machine balance control.In addition, specific experimental tests will be conducted to evaluate the assistive performance of SuperLegs under different working conditions.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.

Figure 1 .
Figure 1.Diagram of a subject walking with weight wearing the SuperLegs.

Figure 2 .
Figure 2. Three configurations of SuperLegs and their corresponding ADLs.a) The convex hexagonal configuration with joints 3 and 6 slides on the ground for walking on flat ground.b) The convex hexagonal configuration cooperated with the human feet alternately for walking on rough ground.c) The pentagonal configuration for stair climbing.d) The concave hexagonal configuration for crouching scene.
the joint position and torque states of SuperLegs, respectively; the output forces at the end are represented by F x , F y , and M z , respectively.The Jacobian matrix of (a)

Figure 4 . 2 dFigure 3 .
Figure 4. a) The load capacity of concave hexagonal configuration.b) The load capacity of convex hexagonal configuration.c) The load capacity of pentagonal configuration.

Figure 7 .
Figure 7.The convergence process of the objective function.

Figure 8 .
Figure 8.Output force results before and after optimization under different configurations.The color represents the maximum output force in the vertical direction.a) Output force results before and after optimization of convex hexagonal configuration.b) Output force results before and after optimization of pentagon configuration.c) Output force results before and after optimization of concave hexagonal configuration.

Figure 9 .
Figure 9. 3D model of the SuperLegs system.a) Detailed view of active joint 1. b) Detailed view of active joint 2. c) Detailed view of active joint 3.

Figure 10 .
Figure 10.The implemented SuperLegs.a) Specific demonstration of unilateral SuperLegs.b) Wearing effect for convex hexagonal configuration.c) Wearing effect for pentagonal configuration.d) Wearing effect for concave hexagonal configuration.

Figure 11 .
Figure 11.a) Experimental setup of different auxiliary configurations.b) The tested and simulated results of different auxiliary configurations.

Figure 12 .
Figure 12.The wearing time of different subjects.

Table 1 .
Parameters of link length used for early-stage design and optimization.
Figure 6.Results of objective function while adjusting dimensions of hip link and foot pedal separately.The other links remain constant during the calculation.

Table 2 .
The weight and load capacity of other SRLGs.