- Open Access
A biologically-inspired multi-joint soft exosuit that can reduce the energy cost of loaded walking
© Panizzolo et al. 2016
- Received: 11 December 2015
- Accepted: 17 April 2016
- Published: 12 May 2016
Carrying load alters normal walking, imposes additional stress to the musculoskeletal system, and results in an increase in energy consumption and a consequent earlier onset of fatigue. This phenomenon is largely due to increased work requirements in lower extremity joints, in turn requiring higher muscle activation. The aim of this work was to assess the biomechanical and physiological effects of a multi-joint soft exosuit that applies assistive torques to the biological hip and ankle joints during loaded walking.
The exosuit was evaluated under three conditions: powered (EXO_ON), unpowered (EXO_OFF) and unpowered removing the equivalent mass of the device (EXO_OFF_EMR). Seven participants walked on an instrumented split-belt treadmill and carried a load equivalent to 30 % their body mass. We assessed their metabolic cost of walking, kinetics, kinematics, and lower limb muscle activation using a portable gas analysis system, motion capture system, and surface electromyography.
Our results showed that the exosuit could deliver controlled forces to a wearer. Net metabolic power in the EXO_ON condition (7.5 ± 0.6 W kg−1) was 7.3 ± 5.0 % and 14.2 ± 6.1 % lower than in the EXO_OFF_EMR condition (7.9 ± 0.8 W kg−1; p = 0.027) and in the EXO_OFF condition (8.5 ± 0.9 W kg−1; p = 0.005), respectively. The exosuit also reduced the total joint positive biological work (sum of hip, knee and ankle) when comparing the EXO_ON condition (1.06 ± 0.16 J kg−1) with respect to the EXO_OFF condition (1.28 ± 0.26 J kg−1; p = 0.020) and to the EXO_OFF_EMR condition (1.22 ± 0.21 J kg−1; p = 0.007).
The results of the present work demonstrate for the first time that a soft wearable robot can improve walking economy. These findings pave the way for future assistive devices that may enhance or restore gait in other applications.
- Soft exosuit
- Metabolic power
- Loaded walking
- Lower limb exoskeleton
Carrying heavy loads alters the biomechanics of walking, leading to an increased metabolic burden. This negative consequence of load carriage has been reported in soldiers, first responders, and recreational athletes who are required to execute physically demanding tasks during walking [1, 2]. Several studies investigating the locomotion of these populations reported increased lower limb joint work [3, 4], which requires higher muscle activation to both sustain the load and stabilize the joints themselves . Higher muscle activity is associated with an increased metabolic cost , leading to an earlier onset of fatigue and an overall reduction of performance [1, 2] while walking. Additionally, prolonged load carriage can result in an increased risk of injury, the most common of which are foot blisters, stress fractures, back strains, metatarsalgia (foot pain), rucksack palsy (shoulder traction injury) and knee pain . Solutions that effectively reduce the burden associated with load carriage during walking are thus warranted.
Lower-limb exoskeletons have been proposed as a means to augment or assist human locomotion for many applications . Some exoskeletons have been designed to make load carriage easier by providing a parallel load path to the ground [8–10], while others apply torques directly to the wearer’s joints [7, 11–14]. These systems are composed of rigid frames that allow the transmission of high forces and, although they represent remarkable achievements, their rigid nature presents a number of practical challenges toward the goal of assisting locomotion. The main challenges arise in aligning the exoskeleton and biological joints with each other  and reducing system mass and in particular distal mass as this can increase metabolic effort .
Our research group has demonstrated reductions in metabolic cost during load carriage with a tethered soft exosuit [20, 24]. One study , conducted with a lab-based, multi-joint tethered actuation platform (composed of a power supply, linear actuators and motor controllers mounted on a stationary platform next to a treadmill), reported reductions in the metabolic cost of walking for hip extension assistance (4.6 %) and for multi-joint assistance (14.6 %). Multi-joint assistance consisted of hip extension, ankle plantarflexion and hip flexion. Though promising, the tethered actuation platform limits the soft exosuit’s applicability to everyday walking.
Therefore, the aim of this work was to perform the first study with an autonomous (fully portable) multi-joint (assisting hip extension, ankle plantarflexion and hip flexion as in ) soft exosuit to evaluate if it could represent an effective solution to reduce the metabolic cost during loaded walking. We evaluated the performance of our soft exosuit on a group of load carriers walking with a load equivalent to 30 % their body weight under three conditions: with the device powered (EXO_ON), with the device unpowered (EXO_OFF) and with the device unpowered with equivalent mass removed (EXO_OFF_EMR). The second condition (EXO_OFF) was evaluated to assess the penalty associated with carrying the additional mass represented by the device itself, an important consideration in the design of such systems. To obtain additional insights on the benefit of wearing the soft exosuit and to extend the knowledge on the biomechanical and physiological effects of this device, we evaluated metabolic cost, muscle activation and joint mechanics which have been shown to be relevant for regulating metabolic energy cost during gait .
Soft exosuit design and operation
The load-transferring elements in the exosuit followed two distinct paths in each leg as described in Fig. 1c. One multiarticular load path extended from the waist, over the front of the thigh, down the side of the leg, passing approximately through the knee joint axis, and to the back of the calf. An additional supporting element attached at the front of the shin. A Bowden cable sheath was anchored to this load path at the back of the calf, with the inner cable extending to the back of the heel. A second monoarticular load path extended from the waist to the back of the hip, and down to the thigh. Along this load path, a Bowden cable sheath connected to the waist belt at the back of the hip, and the inner cable connected to the back of a thigh brace (Fig. 1).
Two actuation units (Fig. 1d) connected to the exosuit via the Bowden cables generated external forces along these two load paths. The multiarticular load path thus enabled the generation of both ankle plantarflexion and hip flexion torques, whereas the monoarticular load path generated hip extension torques.
The ankle and hip contribute together approximately 80 % of the positive mechanical power produced by the lower limb joints during walking [23, 27]. Power generation at the ankle occurs mainly as the body is transitioning from one leg to the other. The ankle on the trailing leg plantarflexes the foot, which propels the body upward and forward to reduce the impact on the leading leg that is accepting the weight of the body. Conversely, power at the hip is generated during extension when it accepts the body’s weight just after heel strike and during flexion when it helps the body to propel forward . This second burst of power generation at the hip occurs as the leg begins to swing forward, which is approximately coincident with the ankle’s pushing off. Therefore, our bioinspired design enables actuation of both joints with a single load path. Details of how the force in the multiarticular load path contributes to hip flexion are explained further in Additional file 1: Text S1.
Soft exosuit actuation and control
The actuation system we built for the exosuit enabled two motors to actuate all four of the load paths (two load paths per leg). Each of the two actuation units (Fig. 1d) consisted of: (1) a module with a geared motor, (2) a multi-wrap pulley connected to two Bowden cables, (3) electronics and (4) a battery module. One actuation unit controlled the bilateral multiarticular load paths, whereas the other actuation unit controlled the bilateral monoarticular load paths as previously described in . In each case, when the motor turned clockwise, the load path on the left leg developed tension while the load path on the right leg was made slack so that no force could be generated. When the motor turned counterclockwise, the load path on the right leg developed tension and the left leg was made slack. Thus, this approach enabled the exosuit to become completely transparent along a particular load path when desired. The total mass of the multi-joint soft exosuit (including the soft exosuit and the two actuation units) was 6.6 kg.
We recruited seven load carriers from the local community (age: 29.3 ± 6.2 yr; height: 1.80 ± 0.07 m; weight: 77.9 ± 8.3 kg). All participants reported no musculoskeletal injuries or diseases and provided written informed consent. The participants whose images appear in the manuscript have provided written consent for the publication of their images according to the policies of the Journal of NeuroEngineering and Rehabilitation. The study was approved by the Harvard Medical School Committee on Human Studies.
Metabolic cost during walking was assessed by indirect calorimetry using a portable gas analysis system (K4b2, Cosmed, Roma, Italy) (Fig. 3b). Carbon dioxide and oxygen rate were averaged across the last two minutes of each condition used to calculate metabolic power using a modified Brockway equation . Net metabolic power for each testing condition was obtained by subtracting the metabolic power obtained during a standing trial performed at the beginning of each session from the metabolic power calculated during the walking conditions. Net metabolic power was normalized by the body mass of each participant.
Joints kinematics and kinetics
Three-dimensional (3D) gait analysis was performed during treadmill walking. The marker set used for 3D motion capture (VICON, Oxford Metrics, UK; 120 Hz) was composed of 50 markers placed on specific anatomical bony landmarks (Fig. 3b-c). Single markers were placed on the left and right legs at the calcanei, heads of the first and fifth metatarsals, medial and lateral malleoli, medial and lateral knee condyles, greater trochanters, left and right anterior superior iliac spines, left and right iliac crests and at the midpoint between the iliac crest and the anterior superior iliac spine on the left and right side. Clusters of four markers were attached to the thighs and shanks of both legs. Eight additional markers were placed on the proximal and distal ends of each cable at the ankle and hip, used to calculate the assistive forces’ lines of action, along with their associated moment arms. All markers and force trajectories were filtered using a zero-lag 4th order low pass Butterworth filter with a 5–9 Hz optimal cut-off frequency selected using a custom residual analysis algorithm (MATLAB, The MathWorks Inc., USA). Joint angles, net joint moments and powers were calculated in the sagittal plane using filtered markers and forces by means of an inverse kinematic and dynamic approach (Visual 3D, C-Motion, Rockville, MD, USA). Net joint moments and powers were then normalized by each participant’s body mass. An automatic gait event detection algorithm (Visual 3D, C-Motion, Rockville, MD, USA) was used to determine heel strikes that defined gait cycles. Ten strides per condition were used for generating mean kinematic and kinetic data for each participant, which were subsequently combined to calculate condition mean data.
During all trials, surface electromyography (EMG) signals from eight lower limb muscles were measured by means of a wired system (Delsys, Natick, MA, USA; 2160 Hz) simultaneously with the motion data measured by the VICON system. Muscles investigated were: rectus femoris (RF), vastus medialis (VM), vastus lateralis (VL), gluteus maximus (GM), biceps femoris (BF), soleus (SOL), medial gastrocnemius (MG) and tibialis anterior (TA) (Fig. 3d). EMG signals were band-pass filtered (4th order Butterworth, cut-off 20–450 Hz), rectified and low-pass filtered (4th order Butterworth, cut-off 6 Hz) to obtain a linear envelope. For each participant and for each muscle, the EMG linear envelope was normalized to the peak value (averaged across ten strides) recorded during the EXO_OFF_EMR condition. Ten strides per condition were used to compute mean muscle activation across each stride.
Biological joint work
Statistical analysis was conducted in SPSS (SPSS Inc., Statistics21, USA). One-way repeated measures analyses of variance (ANOVA) with three modes (EXO_OFF_EMR, EXO_OFF and EXO_ON) were used to verify the effect of the device on metabolic power and on the average muscle activation across the stride. The biomechanical variables of interest included: peak flexion and extension joint angles and moments, as well as absorbed and generated joint power (positive and negative area of the power trace curves) at the ankle, knee and hip. Additional one-way repeated measures ANOVA were conducted on positive and negative biological joint work and power (total and single joint). Bonferroni post hoc tests were performed to identify differences between conditions when a statistically significant main effect was identified by the ANOVA. The significance level was set at p < 0.05 for all analyses. Effect sizes were calculated using Cohen’s d method.
Metabolic cost and muscle activity
Stride length, stride frequency, duty factor and stance and swing times were not significantly different between the three testing conditions. A complete overview of these parameters is presented in Additional file 1: Table S4.
The only statistically significant differences in joint moments (Fig. 7), were a lower peak knee extension moment in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.011; ES = 0.51) and a higher knee flexion moment peak in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.001; ES = 0.30). Significantly higher ankle power generation was reported for the EXO_ON condition with respect to the EXO_OFF_EMR (p = 0.016; ES = 1.45) and to the EXO_OFF (p = 0.005; ES = 1.07) conditions. Additionally, significantly lower knee power absorption was reported in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.003; ES = 0.87) and between the EXO_OFF_EMR condition and the EXO_OFF condition (p = 0.018; ES = 0.59). A complete overview of the joint kinetics data is presented in Additional file 1: Table S5.
Biological joint work and power
Average peak forces generated by the exosuit across the seven participants were 272 ± 43 N of ankle plantarflexion, 204 ± 32 N of hip flexion, and 68 ± 24 N of hip extension. These external forces caused a significant reduction of total joint biological positive work (sum of hip, knee, and ankle) when comparing the EXO_ON condition with respect to the EXO_OFF condition (1.28 ± 0.26 J kg−1; p = 0.020; ES = 1.02) and to the EXO_OFF_EMR condition (1.22 ± 0.21 J kg−1; p = 0.007; ES = 0.86). Further, hip biological positive work was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.011; ES = 1.18) and to the EXO_OFF_EMR condition (p = 0.007; ES = 1.25), and ankle biological positive work was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.035; ES = 0.99). Total joint biological negative work was not significantly different between the three conditions; nevertheless, knee biological negative work was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.003; ES = 0.82).
Similarly, total joint biological positive power (sum of hip, knee, and ankle) was significantly reduced when comparing the EXO_ON (1.02 ± 0.10 W kg−1) with respect to the EXO_OFF condition (1.21 ± 0.19 W kg−1; p = 0.009; ES = 1.25) and to the EXO_OFF_EMR condition (1.16 ± 0.15 W kg−1; p = 0.007; ES = 1.10). Hip biological positive power was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.001; ES = 1.21) and to the EXO_OFF_EMR condition (p = 0.011; ES = 1.21), and ankle biological positive power was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.044; ES = 1.13). Total joint biological negative power was not significantly different between the three conditions. Knee biological negative power was significantly reduced in the EXO_ON condition with respect to the EXO_OFF condition (p = 0.004; ES = 1.20) and in the EXO_OFF condition with respect to the EXO_OFF_EMR condition (p = 0.020; ES = 0.85) (Fig. 5b).
The aim of this study was to evaluate the effects of an autonomous (fully portable) multi-joint soft exosuit on the metabolic cost of loaded walking. A net reduction in the metabolic cost was reported while walking with the exosuit compared to wearing the exoskeleton unpowered with effective mass removed. This finding (Fig. 5a) represents the first successful attempt to effectively reduce the metabolic burden experienced by load carriers with an untethered soft exosuit. In addition, this work also represents the first successful attempt to reduce the metabolic cost of walking with a multi-joint untethered wearable robot of any kind.
Recent work from two different research groups [30, 31] also demonstrated an augmentation of human walking by means of autonomous ankle exoskeletons. These two studies reported average reductions in metabolic cost of 11 ± 4 % and 7.2 ± 2.6 %, respectively, when the ankle joint was assisted during unloaded walking. An additional study  also reported an average reduction of 8 ± 3 % in metabolic cost during loaded walking using a device similar to . Although the magnitude of metabolic reduction observed with the soft exosuit was similar to that reported by these recent studies, differences in our approach to augmenting human performance should be considered when comparing systems and outcomes. Indeed, [30, 32] actively provided assistance at the ankle joint only and delivered higher levels than we did with the soft exosuit at the ankle. This early embodiment of the soft exosuit and actuation units limited the maximum force that could be delivered to the wearer. However, by exploiting the legs being out of phase and timing synergy between hip flexion and ankle plantarflexion we developed an actuation scheme by which a single motor per leg could be used to assist multiple joints, enabling us to minimize system mass and distal mass in particular. Earlier work on multi-joint systems [9, 33, 34] did not report reductions in the metabolic cost of walking, likely due to high device mass.
As expected, the EXO_ON condition produced a larger reduction in metabolic cost when compared to the EXO_OFF condition than to the EXO_OFF_EMR condition. Although the relationship between muscle-tendon behavior and whole body energy consumption is complex , we hypothesized that the underlying physiological mechanisms regulating this interaction could be more pronounced in a EXO_ON vs EXO_OFF comparison and concealed, at least in part, by the additional load imposed by the system in the EXO_OFF_EMR condition.
Walking with the powered device did not alter participants’ spatio-temporal parameters, indicating that the exosuit’s assistive forces were not disruptive to participants’ freely-selected step frequencies and step lengths. Nevertheless, some alterations in participants’ kinematics and kinetics were present. Given the aforementioned finding of reduced metabolic cost during walking, we posit that these changes may have permitted the musculoskeletal system to operate more efficiently. Indeed, considering that ankle dorsiflexion and knee flexion have been shown to increase with an increase of load , our observed reduction of these two parameters in the EXO_ON condition suggests that the exosuit facilitated a return to gait patterns that resemble unloaded walking .
Walking with the soft exosuit reduced the total biological joint work produced by the lower limbs, and the most marked reduction in biological work production was at the hip joint. This reduction is of particular interest especially considering that the hip joint seems to have lower efficiency compared with the other lower limb joints . This hypothesis is also supported by the fact that muscles crossing the hip have a reduced pennation angle and a longer fiber length than those crossing the ankle [36, 37]. This architectural difference, together with a shorter tendon, makes the elastic recoil of the hip muscle-tendon complex less effective compared to that of the ankle . Consequently, work production at the hip during walking is more costly than at the ankle. Based on this rationale and our previous work , it seems possible that unloading the hip joint by means of our multi-joint exosuit could have resulted in a more effective energy saving strategy. Nevertheless, only future experimental studies decoupling the effect of the assistance for each joint could provide further elucidation on this point.
Although the main reduction in the biological work was reported at the hip, the knee and ankle may also have contributed to lowering the metabolic burden. Interestingly, the assistance provided by the soft exosuit reduced knee extension moments (Fig. 7) with an associated reduction in negative biological work. The knee mainly functions as a shock absorber during level ground walking , with the muscles spanning this joint performing mostly eccentric contractions . This behavior is exaggerated during load carriage  and, although this type of contraction is more economical than an isometric or a concentric contraction , it is still associated with a metabolic cost. Therefore, the reduction in negative biological work at the knee may have contributed to a lower metabolic cost as well. Although the multi-articular textile load path does pass approximately through the center of the knee joint, it is possible that a small level of assistance was applied at the knee. However, given this, another hypothesis for the reduced work at the knee related to the assistance of the contralateral ankle joint can be presented, according to . Contralateral ankle assistance is synchronized with the negative work generation at the knee during the gait cycle. An augmented push-off on the contralateral limb could have been beneficial to lower the load at the knee joint. To support this explanation, a higher ankle moment exhibited in the EXO_ON condition during push-off corresponded to a lower knee moment on the contralateral limb during weight acceptance (Fig. 7), similar to what has previously been reported by . Moreover, a similar trend was present in positive ankle work during push-off and in the negative knee work on the contralateral limb during weight acceptance. Decreased production of biological ankle work was also reported in the EXO_ON condition with respect to the EXO_OFF condition. It can be assumed that the ankle joint also contributed to the metabolic reduction and this rationale can be supported by the reduced soleus activation in the EXO_ON condition with respect to the EXO_OFF condition. The reason this was not found when comparing the EXO_ON and EXO_OFF_EMR conditions may have been that the effect was masked by the wearer having to carry the increased mass of the device.
Surprisingly, despite the fact that biological work was significantly reduced, only small differences were found in muscle activation between conditions (Fig. 6). Reported variations of kinematics and kinetics may have been linked to changes in the functional properties of the muscles rather than simply being the result of reduced muscle activation. Although from a joint-level kinematics analysis it is impossible to decouple the individual behaviors of the tendon and muscle, these alterations might have been the result of muscle fascicles working in a more economical region of their force-length relationship. Previous work on an ankle exoskeleton designed for hopping , revealed adaptive changes in the fascicle length of the soleus associated with a reduction in metabolic cost. Nevertheless, at this stage this hypothesis remains speculative and only future work examining in vivo muscle properties could unravel the underlying specific muscle mechanism.
The observed variability in the reduction of metabolic cost between participants may be due to several contributing factors. The fixed external assistance across participants coupled with small variations in how the exosuit fit to different body sizes and shapes resulted in variations in the percentage of the delivered assistance relative to nominal biological joint torques at the hip and ankle. In addition, small variations in how the multi-articular load path crossed the knee may have resulted in small torques at the knee joint for some participants. Additionally, neuromuscular adaptation associated with the use of wearable robots remains an active research area , with little to no work done on loaded walking, and inter-individual adaptations may have contributed to the variation in our findings across subjects.
Based on the results of this study, we believe that there is potential for further enhancing the exosuit’s performance. First, it should be noted that the joint torques applied by the exosuit to the wearer were relatively low compared to the joint moments experienced by load carriers . This was mainly due to the significant compliance in the textile component of the exosuit and its interface to the wearer. In recent work we have demonstrated that higher assistive forces can be delivered that both the ankle and hip with improvements to suit components and actuation units, demonstrating metabolic reductions up to 8.5 and 15 % respectively when assisting only hip extension and ankle plantarflexion with a multi-articular load path similar to that described in this work [42, 43]. These advances are driven by a more rigorous approach to soft exosuit component evaluation and characterization as we describe in [24, 44]. We believe that this will pave the way for future autonomous single and multi-joint soft exosuits that reduce the energy cost of both loaded and unloaded walking. Second, for the multi-joint exosuit presented in this paper, the timing for hip extension assistance was determined using sensor information from the ankle joint of the contralateral limb (as described in Additional file 1: Text S1). This limited the ability to precisely control how assistance was applied during hip extension, as demonstrated by the increased variability in the averaged torque profiles (Fig. 4) both for a given participant and across participants. To address this, we are developing new control approaches for the hip  and ankle  that provide more repeatable forces using inertial sensors located at each individual joint.
Our results demonstrate that an autonomous soft exosuit can reduce the metabolic burden experienced by load carriers, possibly augmenting their overall gait performance. Although many basic fundamental research and development challenges remain in actuator development, textile innovation, sensing and control, this proof of concept study provides the first demonstration of a soft wearable robot to augment gait. This work also presented the first demonstration that an autonomous multi-joint wearable robot can achieve a metabolic reduction. Future studies will be necessary to explore the effects of assisting walking with a soft exosuit and explore if, for a given level of mechanical work, the most effective approach to providing assistance is by augmenting the function of a single joint or multiple joints. Moreover, the realization of devices that assist joints other than the ankle can increase the knowledge on the determinants of energy cost, potentially revealing different adaptive mechanisms of the musculoskeletal system. Finally, apart from assisting load carriers, we are exploring how the soft exosuit can be used as a platform to assist individuals with compromised ability to produce adequate forces during locomotion , paving the way for many translational opportunities of this technology across a range of different populations.
The authors would like to thank Mr. Fabricio Saucedo and Mr. Stephen Allen for their help during data collection and processing, Mr. Ciaran O’Neill for servicing the actuation units and Dr. Louis Awad and Mr. Brendan Quinlivan for editing the manuscript. This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA), Warrior Web Program (Contract No. W911QX-12-C-0084 and W911NF-14-C-0051). The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressly or implied, of DARPA or the U.S. Government. This work was also partially funded by the Wyss Institute for Biologically Inspired Engineering and Harvard John A. Paulson School of Engineering and Applied Sciences.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Griffin TM, Roberts TJ, Kram R. Metabolic cost of generating muscular force in human walking: insights from load-carrying and speed experiments. J Appl Physiol. 2003;95:172–83.View ArticlePubMedGoogle Scholar
- Knapik JJ, Reynolds KL, Harman E. Soldier load carriage: historical, physiological, biomechanical, and medical aspects. Mil Med. 2004;169:45–56.View ArticlePubMedGoogle Scholar
- Huang T-WP, Kuo AD. Mechanics and energetics of load carriage during human walking. J Exp Biol. 2014;217:605–13.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang H, Frame J, Ozimek E, Leib D, Dugan EL. The effects of load carriage and muscle fatigue on lower-extremity joint mechanics. Res Q Exerc Sport. 2013;84:305–12.View ArticlePubMedGoogle Scholar
- Silder A, Delp SL, Besier T. Men and women adopt similar walking mechanics and muscle activation patterns during load carriage. J Biomech. 2013;46:2522–8.View ArticlePubMedGoogle Scholar
- Knapik J, Harman E, Reynolds K. Load carriage using packs: a review of physiological, biomechanical and medical aspects. Appl Ergon. 1996;27:207–16.View ArticlePubMedGoogle Scholar
- Dollar AM, Herr H. Lower extremity exoskeletons and active orthoses: challenges and state-of-the-art. IEEE Trans Robot. 2008;24:144–58.View ArticleGoogle Scholar
- Kazerooni H, Steger R. The Berkeley lower extremity exoskeleton. J Dyn Syst Meas Control. 2006;128:14.View ArticleGoogle Scholar
- Walsh CJ, Endo K, Herr H. A quasi-passive leg exoskeleton for load-carrying augmentation. Int J Humanoid Robot. 2007;4:487–506.View ArticleGoogle Scholar
- Garcia E, Sater J, Main J. Exoskeletons for human performance augmentation (EHPA): A program summary. J Robotics Soc Japan. 2002;20:822–6.View ArticleGoogle Scholar
- Yamamoto K, Ishii M, Hyodo K, Yoshimitsu T, Matsuo T. Development of power assisting suit (miniaturization of supply system to realize wearable suit). JSME Int J Ser C. 2003;46:923–30.View ArticleGoogle Scholar
- Sawicki GS, Ferris DP. Powered ankle exoskeletons reveal the metabolic cost of plantar flexor mechanical work during walking with longer steps at constant step frequency. J Exp Biol. 2009;212:21–31.View ArticlePubMedGoogle Scholar
- Banala SK, Agrawal SK, Scholz SP. Active leg exoskeleton (ALEX) for gait rehabilitation of motor-impaired patients. IEEE Int Conf Rehabil Robot 2007, 401–7Google Scholar
- Kawamoto H, Lee S, Kanbe S, Sankai Y. Power assist method for HAL-3 using emg-based feedback controller. IEEE Int Conf Syst Man Cybern 2003, 1648–53.Google Scholar
- Schiele A. Ergonomics of exoskeletons: Subjective performance metrics. IEEE/RSJ Int Conf Intell Rob and Sys 2009, 480–5Google Scholar
- Browning RC, Modica JR, Kram R, Goswami A. The effects of adding mass to the legs on the energetics and biomechanics of walking. Med Sci Sports Exerc. 2007;39:515–25.View ArticlePubMedGoogle Scholar
- Asbeck AT, De Rossi SMM, Galiana I, Ding Y, Walsh CJ. Stronger, smarter, softer: next-generation wearable robots. IEEE Robot Autom Mag. 2014;21:22–33.View ArticleGoogle Scholar
- Asbeck AT, Dyer RJ, Larusson AF, Walsh CJ. Biologically-inspired soft exosuit. IEEE Int Conf Rehabil Robot 2013, 1–8Google Scholar
- Asbeck AT, Schmidt K, Walsh CJ. Soft exosuit for hip assistance. Rob Auton Syst. 2015;73:102–10.View ArticleGoogle Scholar
- Ding Y, Galiana I, Asbeck A, De Rossi S, Bae J, Santos RT, Araujo VL, Lee S, Holt KG, Walsh C. Biomechanical and physiological evaluation of multi-joint assistance with soft exosuits. IEEE T Neur Sys Reh. 2016;99:1.View ArticleGoogle Scholar
- Ding Y, Galiana I, Asbeck AT, Quinlivan B, De Rossi SMM, Walsh C. Multi-joint actuation platform for lower extremity soft exosuits, IEEE Int Conf Robot Autom 2014, 1327–34Google Scholar
- Asbeck AT, Schmidt K, Galiana I, Walsh C. Multi-joint soft exosuit for gait assistance, IEEE Int Conf Robot Autom 2015, 6197–204Google Scholar
- Farris DJ, Sawicki GS. The mechanics and energetics of human walking and running: a joint level perspective. J R Soc Interface. 2012;9:110–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Asbeck AT, De Rossi S, Holt K, Walsh C. A biologically inspired soft exosuit for walking assistance. Int J Rob Res. 2015;34:744–62.View ArticleGoogle Scholar
- Umberger BR, Rubenson J. Understanding muscle energetics in locomotion: new modeling and experimental approaches. Exerc Sport Sci Rev. 2011;39:59–67.View ArticlePubMedGoogle Scholar
- Zhang J, Cheah CC, Collins SH. Experimental comparison of torque control methods on an ankle exoskeleton during human walking. Int Conf Robot Autom 2015, 5584–9Google Scholar
- Winter DA. The biomechanics and motor control of human gait (University of Waterloo Press. 4th ed. 2009.View ArticleGoogle Scholar
- Brockway JM. Derivation of formulae used to calculate energy expenditure in man. Hum Nutr Clin Nutr. 1987;41:463–71.PubMedGoogle Scholar
- van den Bogert AJ. Exotendons for assistance of human locomotion. Biomed Eng Online. 2003;2:17.View ArticlePubMedPubMed CentralGoogle Scholar
- Mooney LM, Herr HM. Biomechanical walking mechanisms underlying the metabolic reduction caused by an autonomous exoskeleton. J Neuroeng Rehabil. 2016;13:4.View ArticlePubMedPubMed CentralGoogle Scholar
- Collins SH, Wiggin MB, Sawicki GS. Reducing the energy cost of human walking using an unpowered exoskeleton. Nature. 2015;10:15.Google Scholar
- Mooney LM, Rouse EJ, Herr HM. Autonomous exoskeleton reduces metabolic cost of human walking during load carriage. J Neuroeng Rehabil. 2014;11:80.View ArticlePubMedPubMed CentralGoogle Scholar
- Van Dijk W, van der Kooij H, Hekman E. A passive exoskeleton with artificial tendons: design and experimental evaluation. IEEE Int Conf Rehabil Robot 2011, 1–6Google Scholar
- Gregorczyk KN, Hasselquist L, Schiffman JM, Bensel CK, Obusek JP, Gutekunst DJ. Effects of a lower-body exoskeleton device on metabolic cost and gait biomechanics during load carriage. Ergonomics. 2010;53:1263–75.View ArticlePubMedGoogle Scholar
- Sawicki GS, Lewis CL, Ferris DP. It pays to have a spring in your step. Exerc Sport Sci Rev. 2009;37:130–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Kawakami Y, Ichinose Y, Fukunaga T. Architectural and functional features of human triceps surae muscles during contraction. J Appl Physiol. 1998;85:398–404.PubMedGoogle Scholar
- Chleboun GS, France AR, Crill MT, Braddock HK, Howell JN. In vivo measurement of fascicle length and pennation angle of the human biceps femoris muscle. Cells Tissues Organs. 2001;169:401–9.View ArticlePubMedGoogle Scholar
- Lastayo PC, Pierotti DJ, Pifer J, Hoppeler H, Lindstedt SL. Eccentric ergometry: Increases in locomotor muscle size and strength at low training intensities. Am J Physiol Regul Integr Comp Physiol. 2000;278:1282–8.Google Scholar
- Jackson RW, Collins SH. An experimental comparison of the relative benefits of work and torque assistance in ankle exoskeletons. J Appl Physiol. 2015;119:541–57.View ArticlePubMedGoogle Scholar
- Farris DJ, Robertson BD, Sawicki GS. Elastic ankle exoskeletons reduce soleus muscle force but not work in human hopping. J Appl Physiol. 2013;115:579–85.View ArticlePubMedGoogle Scholar
- Reinkensmeyer DJ, Emken JL, Cramer SC. Robotics, motor learning, and neurologic recovery. Annu Rev Biomed Eng. 2004;6:497–525.View ArticlePubMedGoogle Scholar
- Ding Y, Galiana I, Siviy C, Panizzolo FA, Walsh C. IMU-based iterative control for hip extension assistance with a soft exosuit. IEEE Int Conf Robot Autom 2016.Google Scholar
- Lee S, Crea S, Malcolm P, Galiana I, Walsh C. Controlling negative and positive power at the ankle with a soft exosuit. IEEE Int Conf Robot Autom 2016.Google Scholar
- Quinlivan B, Asbeck A, Wagner D, Ranzani T, Russo S, Walsh C. Force transfer characterization of a soft exosuit for gait assistance. ASME Int Design Eng Tech Conf & Comput Info Eng Conf 2015, V05AT08A049.Google Scholar
- Bae J, De Rossi SMM, O’Donnell K, Hendron KL, Awad LN, Teles Dos Santos TR, De Araujo VL, Ding Y, Holt KG, Ellis TD, Walsh CJ. A soft exosuit for patients with stroke: feasibility study with a mobile off-board actuation unit. IEEE Int Conf Rehabil Robot 2015, 131–8.Google Scholar
- Perry J, Burnfield J. Gait analysis: normal and pathological function, Slack Incorporated. 2nd ed. 2010.Google Scholar