Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Lower-Limb-Assisting Robotic Exoskeleton Reduces Energy Consumption in Healthy Young Persons during Stair Climbing

Lower-Limb-Assisting Robotic Exoskeleton Reduces Energy Consumption in Healthy Young Persons... Hindawi Applied Bionics and Biomechanics Volume 2021, Article ID 8833461, 8 pages https://doi.org/10.1155/2021/8833461 Research Article Lower-Limb-Assisting Robotic Exoskeleton Reduces Energy Consumption in Healthy Young Persons during Stair Climbing 1 1,2 2,3 Hanseung Woo , Kyoungchul Kong , and Dong-wook Rha Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Republic of Korea Angel Robotics, Seoul 04798, Republic of Korea Department and Research Institute of Rehabilitation Medicine, Yonsei University College of Medicine, Seoul 03722, Republic of Korea Correspondence should be addressed to Kyoungchul Kong; kckong@kaist.ac.kr and Dong-wook Rha; medicus@yonsei.ac.kr Received 3 July 2020; Accepted 12 April 2021; Published 26 April 2021 Academic Editor: Nan Xiao Copyright © 2021 Hanseung Woo et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Many robotic exoskeletons for lower limb assistance aid walking by reducing energy costs. However, investigations examining stair- climbing assistance have remained limited, generally evaluating reduced activation of related muscles. This study sought to investigate how climbing assistance by a robotic exoskeleton affects energy consumption. Ten healthy young participants wearing a robotic exoskeleton that assists flexion and extension of hip and knee joints walked up nine flights of stairs twice at a self-selected speed with and without stair-climbing assistance. Metabolic cost was assessed by measuring oxygen consumption, heart rate, and the time to climb each flight of stairs. Net oxygen cost (NOC) and total heart beats (THB) were used as measures of metabolic cost, accounting for different climbing speeds. Stair-climbing assistance reduced NOC and THB by 9.3% (P <0:001 ) and 6.9% (P =0:003), respectively, without affecting climbing speed. Despite lack of individual optimization, assistive joint torque applied to the hip and knee joints reduced metabolic cost and cardiovascular burden of stair climbing in healthy young males. These results may be used to improve methods for stair ascent assistance. 1. Introduction pressure source, with the energy cost advantage observed only on a treadmill at a constant walking speed. Various robotic exoskeletons for lower limb assistance have The metabolic cost benefits of the portable autonomous been developed to aid walking, the most common method ankle exoskeleton developed by Mooney et al. [6, 7] were similarly verified in a controlled environment. Recent studies of human locomotion. Investigations of walking mechanics have provided the foundation for the development of such demonstrated that walking assistance by a tethered multi- robots. Kinematics of lower limb joints have been used to joint soft exosuit significantly reduced metabolic costs [8, determine the range of motion and the degree of freedom 9]. Furthermore, an autonomous version of the soft exosuit of robotic joints [1], whereas kinetics have been used to reduced the metabolic cost of walking together with a carry- determine the required robot joint power [2]. Together, kine- ing load [10, 11]. In a case study of overground walking assis- matics and kinetics have provided important insights into the tance with two subjects, the autonomous soft exosuit also development of appropriate walking assistance methods. reduced the metabolic cost of loaded walking over a 500 m The benefits of walking assistance provided by robotic cross-country trail [12]. exoskeletons have been evidenced as reduced energy costs. Although stair climbing is almost as frequent as walking, For instance, ankle exoskeletons utilizing pneumatic muscles the biomechanical characteristics of the two types of motion to assist ankle plantar flexion have been shown to reduce the are distinct from each other. Relative to walking, stair climb- metabolic cost of walking [3–5]. However, these exoskeletons ing, characterized by large joint moment and power, were not fully mobile, as they were powered by an external air increases joint flexion in the hip, knee, and ankle in the 2 Applied Bionics and Biomechanics Control board & battery Signal processing Control algorithm implementation Active hip joint Gait phase detection 1 active DOF (F/E) 2 passive DOF Rotation angle measurement Active knee joint 1 active DOF (F/E) Rotation angle measurement Passive ankle joint 3 passive DOF Air pressure sensors AFO Ground reaction force estimation (Modular Dynamic AFO, Fillauer, Chattanooga, TN, USA) Figure 1: A robotic exoskeleton for lower limb assistance. DOF: degree of freedom; AFO: ankle-foot orthosis; F/E: flexion/extension. sagittal plane. During stair climbing, power generation in the passive ankle joints. Active joints were powered by electrical knee joint is dominant relative to power absorption, and the motors with a gear ratio of 76 : 1, with assistive torque deliv- total positive network of the joints is larger than that during ered to the wearer’s joints in the sagittal plane. Active joints walking [13, 14]. Given that stair climbing requires consider- adopted a series elastic actuation mechanism [21, 22], able joint torque, positive power, and positive network, assis- enabling accurate control of the interaction torque between tance provided by a wearable robot is expected to reduce the robot and human joints. Dynamic ankle-foot orthosis was energetic cost of ascent by providing assistive torque to the used as the passive ankle joint to support the weight of the wearer’s joints. exoskeleton and to provide a sufficient degree of freedom in Previous reports have proposed climbing assistance the wearer’s ankle joint. Foot pressure under the metatarsal methods facilitated by robotic exoskeletons. Despite utilizing joint and the heel was measured using silicon tubes and air different exoskeletons, actuators, sensors, and target users, pressure sensors to estimate the ground reaction force [23]. several studies have verified, using electromyography, the All algorithms required to operate the exoskeleton, such as effects of stair-climbing assistance on subjects capable of per- sensor signal processing and the control algorithm, were forming voluntary leg motions [15–17]. Yet, to date, the implemented using the embedded control board (sbrio effects of exoskeleton-mediated climbing assistance on ener- 9651, National Instruments, TX, USA) and software (Lab- getic costs have not been considered, with a limited number VIEW 2015, National Instruments, TX, USA). The embed- of studies analyzing the metabolic effects of assistance for ded control board and battery, enclosed in a backpack, other related motions. For instance, assistance provided by enabled the exoskeleton to be fully mobile. The total weight a knee exoskeleton during step-up-and-down [18] and squat- of the robotic exoskeleton was 13 kg. ting [19] exercises has been shown to reduce the heart rate and the metabolic equivalent of task [20], respectively. Pres- 2.3. Assistance Strategy for a Stair-Climbing Motion. Com- ently, we sought to investigate the energy consumption effect mon robotic joints are actuated by motors and gear reducers of stair-climbing assistance provided by a robotic exoskele- to amplify output torque to the desired magnitude. However, ton. Specifically, we tested oxygen consumption and heart such a gear train is associated with large resistive torque. To rate during a nine-flight stair climb in the absence or pres- remove the resistance and allow the wearer to move freely ence of assistance from an autonomous (i.e., fully mobile) without discomfort, the robotic exoskeleton used a zero- robotic exoskeleton with rigid frames and braces providing impedance control (ZIC) algorithm [21, 24–26]. As shown assistive torque to the hip and knee joints. in Figure 2, the ZIC application reduced the interaction tor- que between robotic and human joints such that it lies under 0.87 Nm. Also, it allowed the wearer to move the joint much 2. Materials and Methods faster with reduced resistive force. 2.1. Participants. Nondisabled male subjects (N =10, age = Following ZIC-mediated removal of resistance, assistive torque can be provided via an assist-as-needed strategy [27, 28:3±1:7 years, weight = 68:8±5:5 kg, height = 173:8± 5:1 cm) participated in this study. None of the subjects have 28], which requires the determination of the timing for the been diagnosed with musculoskeletal or neurologic disorders application of the assistive torque. Previous studies [5, 29, 30] have used actuation onset timing, with a predefined assis- affecting walking or stair climbing. Ethical approval was granted by the institutional review board and ethics commit- tive force provided at onset timing determined based on the percentages of a gait cycle. Presently, we employed a similar tee (4-2017-0578). strategy for the stair-climbing assistance. 2.2. Robotic Exoskeleton. The robotic exoskeleton used in this An investigation by Riener et al. of stair-climbing biome- study (Figure 1) consisted of active hip and knee joints and chanics and motor coordination [13] revealed substantial Applied Bionics and Biomechanics 3 2.5 1.5 0.5 0 510 15 20 Time (sec.) ZIC on ZIC off (a) –2 –4 05 10 15 20 Time (sec.) ZIC on ZIC off (b) Mechanical impedance of PID controlled SEA –20 0 1 10 10 Frequency (Hz) ZIC on ZIC off (c) Figure 2: Reduced mechanical impedance (resistive force) on the knee joint of the robotic exoskeleton after applying ZIC: (a) an arbitrary motion applied to the joint; (b) a resistive force against motion inputs; (c) a mechanical impedance defined as the magnitude of the ratio of the resistive force to the joint velocity in the frequency domain. extension torque of the hip and knee joints during the early where t , τ , and t indicate onset time, peak value of the i p p section of the stance phase. Flexion torques are also observed assistive torque, and time to reach, τ , respectively. When in both joints during the initial section of the swing phase. τ ðtÞ =0, the desired interaction torque is zero, indicating a The developed assistive joint torque profile for stair climbing zero-impedance mode. Positive and negative values of τ emulated the joint torques during the early sections of the indicate extension and flexion torque, respectively. To pro- stance and swing phases. The assistive torque profile for the vide assistive torque during the early section of both phases, hip and knee joints can be simplified and parameterized as t can be set as the time when each phase is detected by ana- ! ! > lyzing the ground reaction forces [23]. Determination of −0:5 τ cos ðÞ t − t − τ ,if t ∈ t , t +2t , p i p i i p onset timing by phase detection, rather than via percentage τ ðÞ t = p > of gait cycle assessment, allowed the synchronization of assis- 0, otherwise, tive torque with the wearer’s motion regardless of climbing ð1Þ speed changes. To prevent assistive torque for level-ground Angle (rad) Magnitude (dB) Resistive torque (Nm) 4 Applied Bionics and Biomechanics walking during the stair climb, the assistance algorithm ZIC assessed the current ground state. The workflow of stair- climbing assistance is shown in Figure 3. The τ and t , not optimized for each subject, were set p p no On stair? empirically using preliminary experiments (Table 1). The assistive torques defined in the time domain were equally yes applied to all subjects after initiation of motion. However, their application varied in a subject-specific manner based Stair-climbing assistance on stair-climbing speed and stance cycles (Figure 4). no no Stance Swing 2.4. Experimental Protocol. The subjects were asked to rest for detected? detected? at least 5 min in a sitting position. Subsequently, oxygen con- sumption and heart rates were measured for 2 min in a stand- yes yes ing position using a portable metabolic system (K4b2, Extension assistance Flexion assistance COSMED, Rome, Italy) and a heart rate monitor (TICKR (hip & knee) (hip & knee) X, Wahoo, Atlanta, United States), respectively. Following the establishment of the resting heart rate (±5 beats per Figure 3: Workflow schematic of the stair-climbing assistance provided by the robotic exoskeleton. ZIC: zero-impedance control. min), the subjects climbed nine flights of stairs (total of 195 steps; 20 steps per level through the 7th floor and 25 steps Table 1: Defined assistive torque parameters. τ : peak value of the per level from the 7th floor to the 10th floor) twice at a self- p assistive torque; t : time taken to reach the τ . selected speed. One trial was carried out without assistance p p (i.e., zero-impedance mode), with stair-climbing assistance Stance phase Swing phase applied for the other trial, and the order of trials was random- Joint τ t τ t p p p p ized. Subjects rested for at least 10 min between trials. During Hip -27 Nm 0.25 s 16 Nm 0.2 s climbing, oxygen consumption, heart rate, and time to climb Knee -29 Nm 0.25 s 12 Nm 0.2 s each level were measured. A practice period was provided to help subjects adapt to the wearable robot, with a maximum of two practice periods allowed. 2.5. Data Processing. Acustom m-file (MATLAB, Math- Works, Natick, MA, USA) was used for data processing. Oxy- 20 gen consumption and heart rate values were smoothed to 15 1.32 s eliminate measurement noise and outliers using smooth.m, 1.16 s with span = 10%,and method = rloess, where span is the 1.00 s length of the moving window as a percentage of the raw data, with the rloess method applying a second-order regression to the raw data from which the outlier is excluded [31]. Net oxy- gen cost (NOC) was calculated as the difference between the 0 102030405060708090 100 oxygen consumption rate and the average resting oxygen con- Stance cycle (%) sumption rate divided by the stair-climbing speed [32]. Total heart beats (THB) [33] during the climbing session were cal- AVG - STD culated by numerical integration of the heart rate. The average AVG AVG + STD climbing speed was calculated by dividing the number of steps by the climbing time. NOC, THB, and average climbing speed Figure 4: Example of knee extension torque for different stance of the two climbing trials were compared using a paired t-test, times with the same t (0.25 s). AVG and STD indicate the mean with P values < 0.05 considered statistically significant. The and standard deviation of the stance time of the subject, respectively. paired t-test was performed using SPSS ver. 24 (SPSS Inc., Chicago, IL, USA). 4. Discussion 4.1. Stair-Climbing Speed. In contrast to previous studies uti- 3. Results and Discussion lizing treadmills, the present experiments were conducted 3.1. Results. On average, stair-climbing assistance reduced using nine flights of physical stairs. Despite the lack of a con- trolled experimental environment with strict regulation of NOC and THB by 9.3% (P <0:001) and 6.9% (P =0:003), respectively (Figure 5). For all subjects, although the average climbing speed and motion analysis, these results more closely speed per level varied during climbing, no significant reflect real-life circumstances. With the ability to self-select climbing speed, the subjects demonstrated individual and between-trial differences were observed (Figure 6). Knee extension torque (Nm) Applied Bionics and Biomechanics 5 P < 0.001 P = 0.003 0.5 0.45 0.4 0.35 0.3 0.25 0.2 ZIC ASSIST ZIC ASSIST (a) (b) Figure 5: Stair-climbing net oxygen cost (NOC) and total heart beats (THB). NOC (a) and THB (b) were calculated for a 9-level stair climb with (ASSIST) and without climbing assistance (ZIC). 2345678 910 Level (a) 2345678 910 Level (b) Figure 6: Average climbing speed. Climbing speed was measured under zero-impedance control (a) or climbing assistance (b) conditions. Net oxygen cost (ml/kg/step) Stair-climbing speed (steps/min) Stair-climbing speed (steps/min) Total heart beats (beats) 6 Applied Bionics and Biomechanics drawback, this study demonstrated, for the first time, reduc- within-climb differences in ascent speeds. However, the ZIC and assist conditions did not result in significant differences tion in climbing-associated energy consumption using a in the average stair-climbing speeds. These results indicated THKAF-type exoskeleton. Presently, equal assistive torque was provided to all sub- that, under the conditions tested, stair-climbing assistance has no effect on climbing speed. jects, with parameters (Table 1) empirically determined using preliminary studies. However, subject-specific assistive tor- 4.2. NOC Reduction. Multiple studies of exoskeleton assis- que parameters may maximize the assistance effect for tance [3–11], utilizing a treadmill environment to regulate individual subjects. Recent studies have demonstrated that optimization of individual assistive force profiles can further walking speed, evaluated steady-state values of oxygen con- sumption and heart rate as indices of metabolic cost. Pres- reduce energy consumption [11, 36]. Thus, subject-specific ently, we also considered the self-selected climbing speed optimization of the assistive torque provided by the exoskel- for the assessment of metabolic cost. Plasschaert et al. previ- eton tested here may result in improved climbing assistance ously defined NOC as the difference between the average effects. However, despite identical assistive torques, the pres- ent study demonstrated effective stair-climbing assistance for oxygen consumption rate during walking and that during resting divided by the average walking speed [32]. Substitut- all subjects. ing the average climbing speed (steps/min) into the equation, Multiple studies have compared metabolic costs between we found that the metabolic cost decreased by 9.3% in the robotic assistance and no-robot or unpowered conditions [5, presence of stair-climbing assistance. 8, 9, 36, 37]. However, the use of nondisabled subjects profi- cient in performing the tested physical tasks unassisted can 4.3. THB Reduction. In a manner similar to NOC, the heart mask the effects of assistance. The assistance effect can only rate should also be analyzed in a nonsteady state induced emerge when the assistance overcomes the negative effects by climbing speed changes. Hood et al. defined the THB on energy consumption of factors such as exoskeleton index as the ratio of THB during the exercise period to the weight, robot-imposed limitations in joint degree of freedom, and resistive forces of robotic joints [35]. Ding and colleagues total traveled distance [33]. As the subjects covered the same climbing distance in the present study, unadjusted THB [8, 36] used a tethered actuation system to facilitate assis- values were considered. In addition to serving as an accurate tance without increasing the weight load on the subjects. In and convenient estimate of energy expenditure, the heart rate cases where power sources could not be separated from the also directly reflects the cardiac burden of a person. Pres- robot body, unpowered conditions have been used as con- trols to identify the assistance effect [5, 37]. However, it is ently, THB decreased by 6.9% following robotic exoskeleton assistance, suggesting that climbing assistance can reduce also necessary to guarantee compliance of the robotic joints cardiovascular burden. using specific mechanisms or control algorithms. Presently, as the actuation parts of the exoskeleton could not be 4.4. Methodological Issues and Limitations. The present study separated from the robot body, ZIC was applied to ensure compliance of the robotic joints, with the metabolic cost used a trunk-hip-knee-ankle-foot- (THKAF-) type multi- joint robotic exoskeleton [34]. Exoskeletons of this type have compared to that under conditions of applied assistance. been used to provide full assistance to paraplegic patients to Thus, the possibility that the exoskeleton increases metabolic costs relative to no-robot conditions represents another move their legs. Using such exoskeletons on nondisabled wearers requires the exoskeletal joints to have zero imped- limitation of this study. Such limitations, reported in other studies [3, 4], can be addressed by reducing exoskeleton ance, or zero torque [35]. Ankle exoskeletons for walking assistance have provided energetic cost advantages due to weight, improving wearability, and ensuring greater degree negligible device weight and minimal interference in joint of freedom of leg joints. function. Multijoint soft exosuits, utilizing special actuation and power transmission mechanisms, have also considerably 5. Conclusions reduced metabolic costs due to limited weight and minimized resistive force. During stair climbing, the function of the knee Stair-climbing assistance provided by an exoskeleton that joint (i.e., torque, power, and work) supersedes that of other assists flexion and extension of hip and knee joints reduced joints [14], with most exoskeletons for stair-climbing the metabolic cost and cardiovascular burden in healthy assistance previously developed as knee-type devices. Thus, young males climbing nine floors of stairs. stair-climbing assistance using a THKAF-type exoskeleton was expected to be effective for metabolic cost reduction. Data Availability Despite its considerable weight, the THKAF-type exo- skeleton was sufficiently portable to perform experiments The data used to support the findings of this study are avail- using a physical stair environment. However, our experi- able from the corresponding author upon request. ments were limited in their ability to obtain data for biomechanical analyses, including measurements of muscle activity and of human motion in a three-dimensional space. Conflicts of Interest Lack of kinematic and kinetic information, as well as of elec- tromyography data, represents a limitation of this study that The authors declare that there is no conflict of interest hinders direct comparisons with other works. Despite this regarding the publication of this paper. Applied Bionics and Biomechanics 7 [15] C. Fleischer and G. Hommel, “A human–exoskeleton interface Acknowledgments utilizing electromyography,” IEEE Transactions on Robotics, This research was supported by Basic Science Research Program vol. 24, no. 4, pp. 872–882, 2008. through the National Research Foundation of Korea (NRF) [16] T.-J. Yeh, M.-J. Wu, T.-J. Lu, F.-K. Wu, and C.-R. Huang, funded by the Ministry of Education (2016R1D1A1B03934338 “Control of McKibben pneumatic muscles for a power-assist, and 2019R1I1A1A01064032). lower-limb orthosis,” Mechatronics, vol. 20, no. 6, pp. 686– 697, 2010. [17] D. Sasaki, T. Noritsugu, and M. Takaiwa, “Development of References pneumatic lower limb power assist wear driven with wearable air supply system,” in 2013 IEEE/RSJ International Conference [1] A. Chu, H. Kazerooni, and A. Zoss, “On the biomimetic design on Intelligent Robots and Systems, pp. 4440–4445, Tokyo, of the berkeley lower extremity exoskeleton (BLEEX),” in Pro- Japan, 2013. ceedings of the 2005 IEEE International Conference on Robotics [18] T. Nakamura, K. Saito, and K. Kosuge, “Control of wearable and Automation, pp. 4345–4352, Barcelona, Spain, 2005. walking support system based on human-model and grf,” in [2] M. Cenciarini and A. M. Dollar, “Biomechanical consider- Proceedings of the 2005 IEEE International Conference on ations in the design of lower limb exoskeletons,” in Proceedings Robotics and Automation, pp. 4394–4399, Barcelona, Spain, of the IEEE International Conference on Rehabilitation Robot- ics ICORR, pp. 297–302, Zurich, Switzerland, 2011. [19] A. Gams, T. Petriˇc, T. Debevec, and J. Babiˇc, “Effects of [3] J. A. Norris, K. P. Granata, M. R. Mitros, E. M. Byrne, and A. P. robotic knee exoskeleton on human energy expenditure,” IEEE Marsh, “Effect of augmented plantarflexion power on pre- Transactions on Biomedical Engineering, vol. 60, no. 6, ferred walking speed and economy in young and older adults,” pp. 1636–1644, 2013. Gait & Posture, vol. 25, no. 4, pp. 620–627, 2007. [20] B. E. Ainsworth, W. L. Haskell, S. D. Herrmann et al., “2011 [4] G. S. Sawicki and D. P. Ferris, “Mechanics and energetics of compendium of physical activities: a second update of codes level walking with powered ankle exoskeletons,” Journal of and met values,” Medicine & Science in Sports & Exercise, Experimental Biology, vol. 211, no. 9, pp. 1402–1413, 2008. vol. 43, no. 8, pp. 1575–1581, 2011. [5] P. Malcolm, W. Derave, S. Galle, and D. De Clercq, “A simple [21] K. Kong, J. Bae, and M. Tomizuka, “Control of rotary series exoskeleton that assists plantarflexion can reduce the meta- elastic actuator for ideal force-mode actuation in human– bolic cost of human walking,” Plo S one, vol. 8, no. 2, article robot interaction applications,” IEEE/ASME Transactions on e56137, 2013. Mechatronics, vol. 14, no. 1, pp. 105–118, 2009. [6] L. M. Mooney, E. J. Rouse, and H. M. Herr, “Autonomous exo- [22] K. Kong, J. Bae, and M. Tomizuka, “A compact rotary series skeleton reduces metabolic cost of human walking during load elastic actuator for human assistive systems,” IEEE/ASME carriage,” Journal of Neuroengineering and Rehabilitation, Transactions on Mechatronics, vol. 17, no. 2, pp. 288–297, vol. 11, no. 1, p. 80, 2014. [7] L. M. Mooney and H. M. Herr, “Biomechanical walking mech- [23] K. Kong and M. Tomizuka, “Agait monitoring system based on anisms underlying the metabolic reduction caused by an air pressure sensors embedded in a shoe,” IEEE/ASME Transac- autonomous exoskeleton,” Journal of Neuroengineering and tions on Mechatronics,vol. 14,no.3, pp.358–370, 2009. Rehabilitation, vol. 13, no. 1, p. 4, 2016. [24] J. A. Blaya and H. Herr, “Adaptive control of a variable- [8] Y. Ding, I. Galiana, A. T. Asbeck et al., “Biomechanical and impedance ankle-foot orthosis to assist drop-foot gait,” IEEE physiological evaluation of multi-joint assistance with soft Transactions on Neural Systems and Rehabilitation Engineer- exosuits,” IEEE Transactions on Neural Systems and Rehabili- ing, vol. 12, no. 1, pp. 24–31, 2004. tation Engineering, vol. 25, no. 2, pp. 119–130, 2017. [25] J. F. Veneman, R. Ekkelenkamp, R. Kruidhof, F. C. van der [9] B. Quinlivan, S. Lee, P. Malcolm et al., “Assistance magnitude Helm, and H. van der Kooij, “A series elastic-and Bowden- versus metabolic cost reductions for a tethered multiarticular soft cable-based actuation system for use as torque actuator in exosuit,” Science robotics, vol. 2, no. 2, article eaah4416, 2017. exoskeleton-type robots,” The international journal of robotics [10] F. A. Panizzolo, I. Galiana, A. T. Asbeck et al., “A biologically- research, vol. 25, no. 3, pp. 261–281, 2006. inspired multi-joint soft exosuit that can reduce the energy [26] S. Kim and J. Bae, “Force-mode control of rotary series elastic cost of loaded walking,” Journal of Neuro Engineering and actuators in a lower extremity exoskeleton using model- Rehabilitation, vol. 13, no. 1, p. 43, 2016. inverse time delay control,” IEEE/ASME Transactions on [11] S. Lee, J. Kim, L. Baker et al., “Autonomous multi-joint soft Mechatronics, vol. 22, no. 3, pp. 1392–1400, 2017. exosuit with augmentation-power-based control parameter tuning reduces energy cost of loaded walking,” Journal of Neu- [27] S. K. Banala, S. H. Kim, S. K. Agrawal, and J. P. Scholz, “Robot roengineering and Rehabilitation, vol. 15, no. 1, p. 66, 2018. assisted gait training with active leg exoskeleton (alex),” IEEE Transactions on Neural Systems and Rehabilitation Engineer- [12] S. Lee, N. Karavas, B. T. Quinlivan et al., “Autonomous multi- ing, vol. 17, no. 1, pp. 2–8, 2009. joint soft exosuit for assistance with walking overground,” in 2018 IEEE International Conference on Robotics and Automa- [28] E. H. Van Asseldonk, J. F. Veneman, R. Ekkelenkamp, J. H. tion (ICRA), pp. 2812–2819, Brisbane, QLD, Australia, 2018. Buurke, F. C. Van der Helm, and H. van der Kooij, “The effects on kinematics and muscle activity of walking in a robotic gait [13] R. Riener, M. Rabuffetti, and C. Frigo, “Stair ascent and trainer during zero-force control,” IEEE Transactions on Neu- descent at different inclinations,” Gait & Posture, vol. 15, ral Systems and Rehabilitation Engineering, vol. 16, no. 4, no. 1, pp. 32–44, 2002. pp. 360–370, 2008. [14] P. DeVita, J. Helseth, and T. Hortobagyi, “Muscles do more positive than negative work in human locomotion,” Journal [29] Y. Ding, F. A. Panizzolo, C. Siviy et al., “Effect of timing of hip of Experimental Biology, vol. 210, no. 19, pp. 3361–3373, 2007. extension assistance during loaded walking with a soft 8 Applied Bionics and Biomechanics exosuit,” Journal of Neuroengineering and Rehabilitation, vol. 13, no. 1, p. 87, 2016. [30] S. Galle, P. Malcolm, S. H. Collins, and D. De Clercq, “Reduc- ing the metabolic cost of walking with an ankle exoskeleton: interaction between actuation timing and power,” Journal of Neuroengineering and Rehabilitation, vol. 14, no. 1, p. 35, [31] Mathworks Inc, “Matlab smooth function,” Apr. 2018, https:// www.mathworks.com/help/curvefit/smooth.html. [32] F. Plasschaert, K. Jones, and M. Forward, “Energy cost of walking: solving the paradox of steady state in the presence of variable walking speed,” Gait & Posture, vol. 29, no. 2, pp. 311–316, 2009. [33] V. L. Hood, M. H. Granat, D. J. Maxwell, and J. P. Hasler, “A new method of using heart rate to represent energy expendi- ture: the total heart beat index,” Archives of Physical Medicine and Rehabilitation, vol. 83, no. 9, pp. 1266–1273, 2002. [34] T. Yan, M. Cempini, C. M. Oddo, and N. Vitiello, “Review of assistive strategies in powered lower-limb orthoses and exo- skeletons,” Robotics and Autonomous Systems, vol. 64, pp. 120–136, 2015. [35] C. J. Walsh, K. Endo, and H. Herr, “A quasi-passive leg exo- skeleton for load-carrying augmentation,” International Jour- nal of Humanoid Robotics, vol. 4, no. 3, pp. 487–506, 2007. [36] Y. Ding, M. Kim, S. Kuindersma, and C. J. Walsh, “Human-in- the-loop optimization of hip assistance with a soft exosuit dur- ing walking,” Science robotics, vol. 3, no. 15, article eaar5438, [37] S. Galle, P. Malcolm, W. Derave, and D. De Clercq, “Uphill walking with a simple exoskeleton: plantarflexion assistance leads to proximal adaptations,” Gait & Posture, vol. 41, no. 1, pp. 246–251, 2015. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Bionics and Biomechanics Hindawi Publishing Corporation

Lower-Limb-Assisting Robotic Exoskeleton Reduces Energy Consumption in Healthy Young Persons during Stair Climbing

Loading next page...
 
/lp/hindawi-publishing-corporation/lower-limb-assisting-robotic-exoskeleton-reduces-energy-consumption-in-uMmZ4OxY3m
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2021 Hanseung Woo et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1176-2322
eISSN
1754-2103
DOI
10.1155/2021/8833461
Publisher site
See Article on Publisher Site

Abstract

Hindawi Applied Bionics and Biomechanics Volume 2021, Article ID 8833461, 8 pages https://doi.org/10.1155/2021/8833461 Research Article Lower-Limb-Assisting Robotic Exoskeleton Reduces Energy Consumption in Healthy Young Persons during Stair Climbing 1 1,2 2,3 Hanseung Woo , Kyoungchul Kong , and Dong-wook Rha Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Republic of Korea Angel Robotics, Seoul 04798, Republic of Korea Department and Research Institute of Rehabilitation Medicine, Yonsei University College of Medicine, Seoul 03722, Republic of Korea Correspondence should be addressed to Kyoungchul Kong; kckong@kaist.ac.kr and Dong-wook Rha; medicus@yonsei.ac.kr Received 3 July 2020; Accepted 12 April 2021; Published 26 April 2021 Academic Editor: Nan Xiao Copyright © 2021 Hanseung Woo et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Many robotic exoskeletons for lower limb assistance aid walking by reducing energy costs. However, investigations examining stair- climbing assistance have remained limited, generally evaluating reduced activation of related muscles. This study sought to investigate how climbing assistance by a robotic exoskeleton affects energy consumption. Ten healthy young participants wearing a robotic exoskeleton that assists flexion and extension of hip and knee joints walked up nine flights of stairs twice at a self-selected speed with and without stair-climbing assistance. Metabolic cost was assessed by measuring oxygen consumption, heart rate, and the time to climb each flight of stairs. Net oxygen cost (NOC) and total heart beats (THB) were used as measures of metabolic cost, accounting for different climbing speeds. Stair-climbing assistance reduced NOC and THB by 9.3% (P <0:001 ) and 6.9% (P =0:003), respectively, without affecting climbing speed. Despite lack of individual optimization, assistive joint torque applied to the hip and knee joints reduced metabolic cost and cardiovascular burden of stair climbing in healthy young males. These results may be used to improve methods for stair ascent assistance. 1. Introduction pressure source, with the energy cost advantage observed only on a treadmill at a constant walking speed. Various robotic exoskeletons for lower limb assistance have The metabolic cost benefits of the portable autonomous been developed to aid walking, the most common method ankle exoskeleton developed by Mooney et al. [6, 7] were similarly verified in a controlled environment. Recent studies of human locomotion. Investigations of walking mechanics have provided the foundation for the development of such demonstrated that walking assistance by a tethered multi- robots. Kinematics of lower limb joints have been used to joint soft exosuit significantly reduced metabolic costs [8, determine the range of motion and the degree of freedom 9]. Furthermore, an autonomous version of the soft exosuit of robotic joints [1], whereas kinetics have been used to reduced the metabolic cost of walking together with a carry- determine the required robot joint power [2]. Together, kine- ing load [10, 11]. In a case study of overground walking assis- matics and kinetics have provided important insights into the tance with two subjects, the autonomous soft exosuit also development of appropriate walking assistance methods. reduced the metabolic cost of loaded walking over a 500 m The benefits of walking assistance provided by robotic cross-country trail [12]. exoskeletons have been evidenced as reduced energy costs. Although stair climbing is almost as frequent as walking, For instance, ankle exoskeletons utilizing pneumatic muscles the biomechanical characteristics of the two types of motion to assist ankle plantar flexion have been shown to reduce the are distinct from each other. Relative to walking, stair climb- metabolic cost of walking [3–5]. However, these exoskeletons ing, characterized by large joint moment and power, were not fully mobile, as they were powered by an external air increases joint flexion in the hip, knee, and ankle in the 2 Applied Bionics and Biomechanics Control board & battery Signal processing Control algorithm implementation Active hip joint Gait phase detection 1 active DOF (F/E) 2 passive DOF Rotation angle measurement Active knee joint 1 active DOF (F/E) Rotation angle measurement Passive ankle joint 3 passive DOF Air pressure sensors AFO Ground reaction force estimation (Modular Dynamic AFO, Fillauer, Chattanooga, TN, USA) Figure 1: A robotic exoskeleton for lower limb assistance. DOF: degree of freedom; AFO: ankle-foot orthosis; F/E: flexion/extension. sagittal plane. During stair climbing, power generation in the passive ankle joints. Active joints were powered by electrical knee joint is dominant relative to power absorption, and the motors with a gear ratio of 76 : 1, with assistive torque deliv- total positive network of the joints is larger than that during ered to the wearer’s joints in the sagittal plane. Active joints walking [13, 14]. Given that stair climbing requires consider- adopted a series elastic actuation mechanism [21, 22], able joint torque, positive power, and positive network, assis- enabling accurate control of the interaction torque between tance provided by a wearable robot is expected to reduce the robot and human joints. Dynamic ankle-foot orthosis was energetic cost of ascent by providing assistive torque to the used as the passive ankle joint to support the weight of the wearer’s joints. exoskeleton and to provide a sufficient degree of freedom in Previous reports have proposed climbing assistance the wearer’s ankle joint. Foot pressure under the metatarsal methods facilitated by robotic exoskeletons. Despite utilizing joint and the heel was measured using silicon tubes and air different exoskeletons, actuators, sensors, and target users, pressure sensors to estimate the ground reaction force [23]. several studies have verified, using electromyography, the All algorithms required to operate the exoskeleton, such as effects of stair-climbing assistance on subjects capable of per- sensor signal processing and the control algorithm, were forming voluntary leg motions [15–17]. Yet, to date, the implemented using the embedded control board (sbrio effects of exoskeleton-mediated climbing assistance on ener- 9651, National Instruments, TX, USA) and software (Lab- getic costs have not been considered, with a limited number VIEW 2015, National Instruments, TX, USA). The embed- of studies analyzing the metabolic effects of assistance for ded control board and battery, enclosed in a backpack, other related motions. For instance, assistance provided by enabled the exoskeleton to be fully mobile. The total weight a knee exoskeleton during step-up-and-down [18] and squat- of the robotic exoskeleton was 13 kg. ting [19] exercises has been shown to reduce the heart rate and the metabolic equivalent of task [20], respectively. Pres- 2.3. Assistance Strategy for a Stair-Climbing Motion. Com- ently, we sought to investigate the energy consumption effect mon robotic joints are actuated by motors and gear reducers of stair-climbing assistance provided by a robotic exoskele- to amplify output torque to the desired magnitude. However, ton. Specifically, we tested oxygen consumption and heart such a gear train is associated with large resistive torque. To rate during a nine-flight stair climb in the absence or pres- remove the resistance and allow the wearer to move freely ence of assistance from an autonomous (i.e., fully mobile) without discomfort, the robotic exoskeleton used a zero- robotic exoskeleton with rigid frames and braces providing impedance control (ZIC) algorithm [21, 24–26]. As shown assistive torque to the hip and knee joints. in Figure 2, the ZIC application reduced the interaction tor- que between robotic and human joints such that it lies under 0.87 Nm. Also, it allowed the wearer to move the joint much 2. Materials and Methods faster with reduced resistive force. 2.1. Participants. Nondisabled male subjects (N =10, age = Following ZIC-mediated removal of resistance, assistive torque can be provided via an assist-as-needed strategy [27, 28:3±1:7 years, weight = 68:8±5:5 kg, height = 173:8± 5:1 cm) participated in this study. None of the subjects have 28], which requires the determination of the timing for the been diagnosed with musculoskeletal or neurologic disorders application of the assistive torque. Previous studies [5, 29, 30] have used actuation onset timing, with a predefined assis- affecting walking or stair climbing. Ethical approval was granted by the institutional review board and ethics commit- tive force provided at onset timing determined based on the percentages of a gait cycle. Presently, we employed a similar tee (4-2017-0578). strategy for the stair-climbing assistance. 2.2. Robotic Exoskeleton. The robotic exoskeleton used in this An investigation by Riener et al. of stair-climbing biome- study (Figure 1) consisted of active hip and knee joints and chanics and motor coordination [13] revealed substantial Applied Bionics and Biomechanics 3 2.5 1.5 0.5 0 510 15 20 Time (sec.) ZIC on ZIC off (a) –2 –4 05 10 15 20 Time (sec.) ZIC on ZIC off (b) Mechanical impedance of PID controlled SEA –20 0 1 10 10 Frequency (Hz) ZIC on ZIC off (c) Figure 2: Reduced mechanical impedance (resistive force) on the knee joint of the robotic exoskeleton after applying ZIC: (a) an arbitrary motion applied to the joint; (b) a resistive force against motion inputs; (c) a mechanical impedance defined as the magnitude of the ratio of the resistive force to the joint velocity in the frequency domain. extension torque of the hip and knee joints during the early where t , τ , and t indicate onset time, peak value of the i p p section of the stance phase. Flexion torques are also observed assistive torque, and time to reach, τ , respectively. When in both joints during the initial section of the swing phase. τ ðtÞ =0, the desired interaction torque is zero, indicating a The developed assistive joint torque profile for stair climbing zero-impedance mode. Positive and negative values of τ emulated the joint torques during the early sections of the indicate extension and flexion torque, respectively. To pro- stance and swing phases. The assistive torque profile for the vide assistive torque during the early section of both phases, hip and knee joints can be simplified and parameterized as t can be set as the time when each phase is detected by ana- ! ! > lyzing the ground reaction forces [23]. Determination of −0:5 τ cos ðÞ t − t − τ ,if t ∈ t , t +2t , p i p i i p onset timing by phase detection, rather than via percentage τ ðÞ t = p > of gait cycle assessment, allowed the synchronization of assis- 0, otherwise, tive torque with the wearer’s motion regardless of climbing ð1Þ speed changes. To prevent assistive torque for level-ground Angle (rad) Magnitude (dB) Resistive torque (Nm) 4 Applied Bionics and Biomechanics walking during the stair climb, the assistance algorithm ZIC assessed the current ground state. The workflow of stair- climbing assistance is shown in Figure 3. The τ and t , not optimized for each subject, were set p p no On stair? empirically using preliminary experiments (Table 1). The assistive torques defined in the time domain were equally yes applied to all subjects after initiation of motion. However, their application varied in a subject-specific manner based Stair-climbing assistance on stair-climbing speed and stance cycles (Figure 4). no no Stance Swing 2.4. Experimental Protocol. The subjects were asked to rest for detected? detected? at least 5 min in a sitting position. Subsequently, oxygen con- sumption and heart rates were measured for 2 min in a stand- yes yes ing position using a portable metabolic system (K4b2, Extension assistance Flexion assistance COSMED, Rome, Italy) and a heart rate monitor (TICKR (hip & knee) (hip & knee) X, Wahoo, Atlanta, United States), respectively. Following the establishment of the resting heart rate (±5 beats per Figure 3: Workflow schematic of the stair-climbing assistance provided by the robotic exoskeleton. ZIC: zero-impedance control. min), the subjects climbed nine flights of stairs (total of 195 steps; 20 steps per level through the 7th floor and 25 steps Table 1: Defined assistive torque parameters. τ : peak value of the per level from the 7th floor to the 10th floor) twice at a self- p assistive torque; t : time taken to reach the τ . selected speed. One trial was carried out without assistance p p (i.e., zero-impedance mode), with stair-climbing assistance Stance phase Swing phase applied for the other trial, and the order of trials was random- Joint τ t τ t p p p p ized. Subjects rested for at least 10 min between trials. During Hip -27 Nm 0.25 s 16 Nm 0.2 s climbing, oxygen consumption, heart rate, and time to climb Knee -29 Nm 0.25 s 12 Nm 0.2 s each level were measured. A practice period was provided to help subjects adapt to the wearable robot, with a maximum of two practice periods allowed. 2.5. Data Processing. Acustom m-file (MATLAB, Math- Works, Natick, MA, USA) was used for data processing. Oxy- 20 gen consumption and heart rate values were smoothed to 15 1.32 s eliminate measurement noise and outliers using smooth.m, 1.16 s with span = 10%,and method = rloess, where span is the 1.00 s length of the moving window as a percentage of the raw data, with the rloess method applying a second-order regression to the raw data from which the outlier is excluded [31]. Net oxy- gen cost (NOC) was calculated as the difference between the 0 102030405060708090 100 oxygen consumption rate and the average resting oxygen con- Stance cycle (%) sumption rate divided by the stair-climbing speed [32]. Total heart beats (THB) [33] during the climbing session were cal- AVG - STD culated by numerical integration of the heart rate. The average AVG AVG + STD climbing speed was calculated by dividing the number of steps by the climbing time. NOC, THB, and average climbing speed Figure 4: Example of knee extension torque for different stance of the two climbing trials were compared using a paired t-test, times with the same t (0.25 s). AVG and STD indicate the mean with P values < 0.05 considered statistically significant. The and standard deviation of the stance time of the subject, respectively. paired t-test was performed using SPSS ver. 24 (SPSS Inc., Chicago, IL, USA). 4. Discussion 4.1. Stair-Climbing Speed. In contrast to previous studies uti- 3. Results and Discussion lizing treadmills, the present experiments were conducted 3.1. Results. On average, stair-climbing assistance reduced using nine flights of physical stairs. Despite the lack of a con- trolled experimental environment with strict regulation of NOC and THB by 9.3% (P <0:001) and 6.9% (P =0:003), respectively (Figure 5). For all subjects, although the average climbing speed and motion analysis, these results more closely speed per level varied during climbing, no significant reflect real-life circumstances. With the ability to self-select climbing speed, the subjects demonstrated individual and between-trial differences were observed (Figure 6). Knee extension torque (Nm) Applied Bionics and Biomechanics 5 P < 0.001 P = 0.003 0.5 0.45 0.4 0.35 0.3 0.25 0.2 ZIC ASSIST ZIC ASSIST (a) (b) Figure 5: Stair-climbing net oxygen cost (NOC) and total heart beats (THB). NOC (a) and THB (b) were calculated for a 9-level stair climb with (ASSIST) and without climbing assistance (ZIC). 2345678 910 Level (a) 2345678 910 Level (b) Figure 6: Average climbing speed. Climbing speed was measured under zero-impedance control (a) or climbing assistance (b) conditions. Net oxygen cost (ml/kg/step) Stair-climbing speed (steps/min) Stair-climbing speed (steps/min) Total heart beats (beats) 6 Applied Bionics and Biomechanics drawback, this study demonstrated, for the first time, reduc- within-climb differences in ascent speeds. However, the ZIC and assist conditions did not result in significant differences tion in climbing-associated energy consumption using a in the average stair-climbing speeds. These results indicated THKAF-type exoskeleton. Presently, equal assistive torque was provided to all sub- that, under the conditions tested, stair-climbing assistance has no effect on climbing speed. jects, with parameters (Table 1) empirically determined using preliminary studies. However, subject-specific assistive tor- 4.2. NOC Reduction. Multiple studies of exoskeleton assis- que parameters may maximize the assistance effect for tance [3–11], utilizing a treadmill environment to regulate individual subjects. Recent studies have demonstrated that optimization of individual assistive force profiles can further walking speed, evaluated steady-state values of oxygen con- sumption and heart rate as indices of metabolic cost. Pres- reduce energy consumption [11, 36]. Thus, subject-specific ently, we also considered the self-selected climbing speed optimization of the assistive torque provided by the exoskel- for the assessment of metabolic cost. Plasschaert et al. previ- eton tested here may result in improved climbing assistance ously defined NOC as the difference between the average effects. However, despite identical assistive torques, the pres- ent study demonstrated effective stair-climbing assistance for oxygen consumption rate during walking and that during resting divided by the average walking speed [32]. Substitut- all subjects. ing the average climbing speed (steps/min) into the equation, Multiple studies have compared metabolic costs between we found that the metabolic cost decreased by 9.3% in the robotic assistance and no-robot or unpowered conditions [5, presence of stair-climbing assistance. 8, 9, 36, 37]. However, the use of nondisabled subjects profi- cient in performing the tested physical tasks unassisted can 4.3. THB Reduction. In a manner similar to NOC, the heart mask the effects of assistance. The assistance effect can only rate should also be analyzed in a nonsteady state induced emerge when the assistance overcomes the negative effects by climbing speed changes. Hood et al. defined the THB on energy consumption of factors such as exoskeleton index as the ratio of THB during the exercise period to the weight, robot-imposed limitations in joint degree of freedom, and resistive forces of robotic joints [35]. Ding and colleagues total traveled distance [33]. As the subjects covered the same climbing distance in the present study, unadjusted THB [8, 36] used a tethered actuation system to facilitate assis- values were considered. In addition to serving as an accurate tance without increasing the weight load on the subjects. In and convenient estimate of energy expenditure, the heart rate cases where power sources could not be separated from the also directly reflects the cardiac burden of a person. Pres- robot body, unpowered conditions have been used as con- trols to identify the assistance effect [5, 37]. However, it is ently, THB decreased by 6.9% following robotic exoskeleton assistance, suggesting that climbing assistance can reduce also necessary to guarantee compliance of the robotic joints cardiovascular burden. using specific mechanisms or control algorithms. Presently, as the actuation parts of the exoskeleton could not be 4.4. Methodological Issues and Limitations. The present study separated from the robot body, ZIC was applied to ensure compliance of the robotic joints, with the metabolic cost used a trunk-hip-knee-ankle-foot- (THKAF-) type multi- joint robotic exoskeleton [34]. Exoskeletons of this type have compared to that under conditions of applied assistance. been used to provide full assistance to paraplegic patients to Thus, the possibility that the exoskeleton increases metabolic costs relative to no-robot conditions represents another move their legs. Using such exoskeletons on nondisabled wearers requires the exoskeletal joints to have zero imped- limitation of this study. Such limitations, reported in other studies [3, 4], can be addressed by reducing exoskeleton ance, or zero torque [35]. Ankle exoskeletons for walking assistance have provided energetic cost advantages due to weight, improving wearability, and ensuring greater degree negligible device weight and minimal interference in joint of freedom of leg joints. function. Multijoint soft exosuits, utilizing special actuation and power transmission mechanisms, have also considerably 5. Conclusions reduced metabolic costs due to limited weight and minimized resistive force. During stair climbing, the function of the knee Stair-climbing assistance provided by an exoskeleton that joint (i.e., torque, power, and work) supersedes that of other assists flexion and extension of hip and knee joints reduced joints [14], with most exoskeletons for stair-climbing the metabolic cost and cardiovascular burden in healthy assistance previously developed as knee-type devices. Thus, young males climbing nine floors of stairs. stair-climbing assistance using a THKAF-type exoskeleton was expected to be effective for metabolic cost reduction. Data Availability Despite its considerable weight, the THKAF-type exo- skeleton was sufficiently portable to perform experiments The data used to support the findings of this study are avail- using a physical stair environment. However, our experi- able from the corresponding author upon request. ments were limited in their ability to obtain data for biomechanical analyses, including measurements of muscle activity and of human motion in a three-dimensional space. Conflicts of Interest Lack of kinematic and kinetic information, as well as of elec- tromyography data, represents a limitation of this study that The authors declare that there is no conflict of interest hinders direct comparisons with other works. Despite this regarding the publication of this paper. Applied Bionics and Biomechanics 7 [15] C. Fleischer and G. Hommel, “A human–exoskeleton interface Acknowledgments utilizing electromyography,” IEEE Transactions on Robotics, This research was supported by Basic Science Research Program vol. 24, no. 4, pp. 872–882, 2008. through the National Research Foundation of Korea (NRF) [16] T.-J. Yeh, M.-J. Wu, T.-J. Lu, F.-K. Wu, and C.-R. Huang, funded by the Ministry of Education (2016R1D1A1B03934338 “Control of McKibben pneumatic muscles for a power-assist, and 2019R1I1A1A01064032). lower-limb orthosis,” Mechatronics, vol. 20, no. 6, pp. 686– 697, 2010. [17] D. Sasaki, T. Noritsugu, and M. Takaiwa, “Development of References pneumatic lower limb power assist wear driven with wearable air supply system,” in 2013 IEEE/RSJ International Conference [1] A. Chu, H. Kazerooni, and A. Zoss, “On the biomimetic design on Intelligent Robots and Systems, pp. 4440–4445, Tokyo, of the berkeley lower extremity exoskeleton (BLEEX),” in Pro- Japan, 2013. ceedings of the 2005 IEEE International Conference on Robotics [18] T. Nakamura, K. Saito, and K. Kosuge, “Control of wearable and Automation, pp. 4345–4352, Barcelona, Spain, 2005. walking support system based on human-model and grf,” in [2] M. Cenciarini and A. M. Dollar, “Biomechanical consider- Proceedings of the 2005 IEEE International Conference on ations in the design of lower limb exoskeletons,” in Proceedings Robotics and Automation, pp. 4394–4399, Barcelona, Spain, of the IEEE International Conference on Rehabilitation Robot- ics ICORR, pp. 297–302, Zurich, Switzerland, 2011. [19] A. Gams, T. Petriˇc, T. Debevec, and J. Babiˇc, “Effects of [3] J. A. Norris, K. P. Granata, M. R. Mitros, E. M. Byrne, and A. P. robotic knee exoskeleton on human energy expenditure,” IEEE Marsh, “Effect of augmented plantarflexion power on pre- Transactions on Biomedical Engineering, vol. 60, no. 6, ferred walking speed and economy in young and older adults,” pp. 1636–1644, 2013. Gait & Posture, vol. 25, no. 4, pp. 620–627, 2007. [20] B. E. Ainsworth, W. L. Haskell, S. D. Herrmann et al., “2011 [4] G. S. Sawicki and D. P. Ferris, “Mechanics and energetics of compendium of physical activities: a second update of codes level walking with powered ankle exoskeletons,” Journal of and met values,” Medicine & Science in Sports & Exercise, Experimental Biology, vol. 211, no. 9, pp. 1402–1413, 2008. vol. 43, no. 8, pp. 1575–1581, 2011. [5] P. Malcolm, W. Derave, S. Galle, and D. De Clercq, “A simple [21] K. Kong, J. Bae, and M. Tomizuka, “Control of rotary series exoskeleton that assists plantarflexion can reduce the meta- elastic actuator for ideal force-mode actuation in human– bolic cost of human walking,” Plo S one, vol. 8, no. 2, article robot interaction applications,” IEEE/ASME Transactions on e56137, 2013. Mechatronics, vol. 14, no. 1, pp. 105–118, 2009. [6] L. M. Mooney, E. J. Rouse, and H. M. Herr, “Autonomous exo- [22] K. Kong, J. Bae, and M. Tomizuka, “A compact rotary series skeleton reduces metabolic cost of human walking during load elastic actuator for human assistive systems,” IEEE/ASME carriage,” Journal of Neuroengineering and Rehabilitation, Transactions on Mechatronics, vol. 17, no. 2, pp. 288–297, vol. 11, no. 1, p. 80, 2014. [7] L. M. Mooney and H. M. Herr, “Biomechanical walking mech- [23] K. Kong and M. Tomizuka, “Agait monitoring system based on anisms underlying the metabolic reduction caused by an air pressure sensors embedded in a shoe,” IEEE/ASME Transac- autonomous exoskeleton,” Journal of Neuroengineering and tions on Mechatronics,vol. 14,no.3, pp.358–370, 2009. Rehabilitation, vol. 13, no. 1, p. 4, 2016. [24] J. A. Blaya and H. Herr, “Adaptive control of a variable- [8] Y. Ding, I. Galiana, A. T. Asbeck et al., “Biomechanical and impedance ankle-foot orthosis to assist drop-foot gait,” IEEE physiological evaluation of multi-joint assistance with soft Transactions on Neural Systems and Rehabilitation Engineer- exosuits,” IEEE Transactions on Neural Systems and Rehabili- ing, vol. 12, no. 1, pp. 24–31, 2004. tation Engineering, vol. 25, no. 2, pp. 119–130, 2017. [25] J. F. Veneman, R. Ekkelenkamp, R. Kruidhof, F. C. van der [9] B. Quinlivan, S. Lee, P. Malcolm et al., “Assistance magnitude Helm, and H. van der Kooij, “A series elastic-and Bowden- versus metabolic cost reductions for a tethered multiarticular soft cable-based actuation system for use as torque actuator in exosuit,” Science robotics, vol. 2, no. 2, article eaah4416, 2017. exoskeleton-type robots,” The international journal of robotics [10] F. A. Panizzolo, I. Galiana, A. T. Asbeck et al., “A biologically- research, vol. 25, no. 3, pp. 261–281, 2006. inspired multi-joint soft exosuit that can reduce the energy [26] S. Kim and J. Bae, “Force-mode control of rotary series elastic cost of loaded walking,” Journal of Neuro Engineering and actuators in a lower extremity exoskeleton using model- Rehabilitation, vol. 13, no. 1, p. 43, 2016. inverse time delay control,” IEEE/ASME Transactions on [11] S. Lee, J. Kim, L. Baker et al., “Autonomous multi-joint soft Mechatronics, vol. 22, no. 3, pp. 1392–1400, 2017. exosuit with augmentation-power-based control parameter tuning reduces energy cost of loaded walking,” Journal of Neu- [27] S. K. Banala, S. H. Kim, S. K. Agrawal, and J. P. Scholz, “Robot roengineering and Rehabilitation, vol. 15, no. 1, p. 66, 2018. assisted gait training with active leg exoskeleton (alex),” IEEE Transactions on Neural Systems and Rehabilitation Engineer- [12] S. Lee, N. Karavas, B. T. Quinlivan et al., “Autonomous multi- ing, vol. 17, no. 1, pp. 2–8, 2009. joint soft exosuit for assistance with walking overground,” in 2018 IEEE International Conference on Robotics and Automa- [28] E. H. Van Asseldonk, J. F. Veneman, R. Ekkelenkamp, J. H. tion (ICRA), pp. 2812–2819, Brisbane, QLD, Australia, 2018. Buurke, F. C. Van der Helm, and H. van der Kooij, “The effects on kinematics and muscle activity of walking in a robotic gait [13] R. Riener, M. Rabuffetti, and C. Frigo, “Stair ascent and trainer during zero-force control,” IEEE Transactions on Neu- descent at different inclinations,” Gait & Posture, vol. 15, ral Systems and Rehabilitation Engineering, vol. 16, no. 4, no. 1, pp. 32–44, 2002. pp. 360–370, 2008. [14] P. DeVita, J. Helseth, and T. Hortobagyi, “Muscles do more positive than negative work in human locomotion,” Journal [29] Y. Ding, F. A. Panizzolo, C. Siviy et al., “Effect of timing of hip of Experimental Biology, vol. 210, no. 19, pp. 3361–3373, 2007. extension assistance during loaded walking with a soft 8 Applied Bionics and Biomechanics exosuit,” Journal of Neuroengineering and Rehabilitation, vol. 13, no. 1, p. 87, 2016. [30] S. Galle, P. Malcolm, S. H. Collins, and D. De Clercq, “Reduc- ing the metabolic cost of walking with an ankle exoskeleton: interaction between actuation timing and power,” Journal of Neuroengineering and Rehabilitation, vol. 14, no. 1, p. 35, [31] Mathworks Inc, “Matlab smooth function,” Apr. 2018, https:// www.mathworks.com/help/curvefit/smooth.html. [32] F. Plasschaert, K. Jones, and M. Forward, “Energy cost of walking: solving the paradox of steady state in the presence of variable walking speed,” Gait & Posture, vol. 29, no. 2, pp. 311–316, 2009. [33] V. L. Hood, M. H. Granat, D. J. Maxwell, and J. P. Hasler, “A new method of using heart rate to represent energy expendi- ture: the total heart beat index,” Archives of Physical Medicine and Rehabilitation, vol. 83, no. 9, pp. 1266–1273, 2002. [34] T. Yan, M. Cempini, C. M. Oddo, and N. Vitiello, “Review of assistive strategies in powered lower-limb orthoses and exo- skeletons,” Robotics and Autonomous Systems, vol. 64, pp. 120–136, 2015. [35] C. J. Walsh, K. Endo, and H. Herr, “A quasi-passive leg exo- skeleton for load-carrying augmentation,” International Jour- nal of Humanoid Robotics, vol. 4, no. 3, pp. 487–506, 2007. [36] Y. Ding, M. Kim, S. Kuindersma, and C. J. Walsh, “Human-in- the-loop optimization of hip assistance with a soft exosuit dur- ing walking,” Science robotics, vol. 3, no. 15, article eaar5438, [37] S. Galle, P. Malcolm, W. Derave, and D. De Clercq, “Uphill walking with a simple exoskeleton: plantarflexion assistance leads to proximal adaptations,” Gait & Posture, vol. 41, no. 1, pp. 246–251, 2015.

Journal

Applied Bionics and BiomechanicsHindawi Publishing Corporation

Published: Apr 26, 2021

References