701

NeuroRehabilitation 35 (2014) 701–709 DOI:10.3233/NRE-141174 IOS Press

State-of-the-art robotic gait rehabilitation orthoses: Design and control aspects Shahid Hussain∗ Department of Robotics and Mechatronics, Nazarbayev University, Astana, Kazakhstan

Abstract. BACKGROUND: Robot assisted gait training is a rapidly evolving rehabilitation practice. Various robotic orthoses have been developed during the past two decades for the gait training of patients suffering from neurologic injuries. These robotic orthoses can provide systematic gait training and reduce the work load of physical therapists. Biomechanical gait parameters can also be recorded and analysed more precisely as compared to manual physical therapy. OBJECTIVES: A review of robotic orthoses developed for providing gait training of neurologically impaired patients is provided in this paper. METHODS: Recent developments in the mechanism design and actuation methods of these robotic gait training orthoses are presented. Control strategies developed for these robotic gait training orthoses in the recent years are also discussed in detail. These control strategies have the capability to provide customised gait training according to the disability level and stage of rehabilitation of neurologically impaired subjects. RESULTS: A detailed discussion regarding the mechanism design, actuation and control strategies with potential developments and improvements is provided at the end of the paper. CONCLUSIONS: A number of robotic orthoses and novel control strategies have been developed to provide gait training according to the disability level of patients and have shown encouraging results. There is a need to develop improved robotic mechanisms, actuation methods and control strategies that can provide naturalistic gait patterns, safe human-robot interaction and customized gait training, respectively. Extensive clinical trials need to be carried out to ascertain the efficacy of these robotic rehabilitation orthoses. Keywords: Control strategies, gait rehabilitation, neurologic injuries, robot desing, robotic orthosis

1. Introduction Lower limb impairments and gait disorders often result from neurologic injuries such as stroke and spinal cord injuries (SCI) (Roger, Lloyd-Jones, & Berry, 2011). In most of the cases these injuries severely affect the survivors’ capabilities to perform activates of daily living (ADL). Since the survival rate of first time stroke victims is more than 50% (Stegall, Winfree, Zanotto, & Agrawal, 2013), these survivors need some form of rehabilitation therapy to regain the required capability ∗ Address for correspondence: Shahid Hussain, Department of Robotics and Mechatronics, Nazarbayev University, Astana, Kazakhstan. Tel.: +7 7172 709105; Fax: +7 7172 70 60 54; E-mail: [email protected].

of performing ADL. Similarly, the improvement in muscular capabilities of incomplete paraplegic patients could be achieved by utilizing rehabilitation therapies (H Barbeau, 1999). Repetitive and intense task specific gait training sessions (French et al., 2007; Kwakkel, Wagenaar, Twisk, Lankhorst, & Koetsier, 1999) may help in improving the motor function of neurologically impaired patients (Cramer & Riley, 2008; Van Peppen et al., 2004). The concept of body weight supported (BWS) treadmill training has been widely used for the physical therapy of patients suffering from stroke and incomplete spinal cord injuries (ISCI) (Barbeau, 1999; Barbeau & Fung, 2001; Da Cunha Jr et al., 2002; Laufer, Dickstein, Chefez, & Marcovitz, 2001; McCain et al., 2008;

1053-8135/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved

702

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

Mulroy et al., 2010). Significant improvements in gait velocity, endurance and step length have been reported after intensive BWS treadmill training (Hassid, Rose, Commisarow, Guttry, & Dobkin, 1997; Hesse et al., 1995; Hesse, Konrad, & Uhlenbrock, 1999; Laufer et al., 2001; Murray, Spurr, & Sepic, 1985; Patterson, Rodgers, Macko, & Forrester, 2008; Teixeira da Cunha Filho et al., 2001; Visintin, Barbeau, KornerBitensky, & Mayo, 1998). A team of physiotherapists is required to guide the limbs and stabilize the pelvis of the patients’ during BWS treadmill training. This results in non-intensive training sessions due to physiotherapists’ fatigue (Holder et al., 1999). The manual physical therapy leaves the judgment of patients’ progress and recovery solely on the physiotherapists’ observations, with no proper method of systematically recording and analysing patients’ recovery. The manual physical therapy also puts more burden on the health care system as the neurologically impaired patients may have to rely on the rehabilitation therapy for prolonged durations (Riener, L¨unenburger, Maier, Colombo, & Dietz, 2010). Robot assisted BWS treadmill training has been introduced in the late 1990’s to overcome the limitations of manual physical therapy (Colombo, Joerg, Schreier, & Dietz, 2000). Robot assisted gait rehabilitation saves the physiotherapists from fatigue and can provide systematic and intensive gait training sessions as compared to manual physiotherapy (Colombo et al., 2000; Hussain, Xie, & Liu, 2011; Pennycott, Wyss, Vallery, Klamroth-Marganska, & Riener). A physiotherapist can supervise a number of patients in the rehabilitation clinics undergoing robot assisted gait training in contrary to a team of physiotherapists who can provide manual physical therapy to a single patient. Further advancements in rehabilitation robotics technology may also reduce the potential cost of gait rehabilitation as compared to manual physiotherapy. Robotic exoskeletons or robotic orthoses are the most commonly used form of devices for providing automated BWS treadmill training (Banala, Kim, Agrawal, & Scholz, 2009; Colombo et al., 2000; Daisuke Aoyagi, 2007; Hussain, Xie, Jamwal, & Parsons, 2012; Moreno et al.; Veneman, Ekkelenkamp, Kruidhof, Van Der Helm, & Van Der Kooij, 2006; Veneman et al., 2007). These wearable robotic exoskeletons work in close proximity to patients and their mechanical joints correspond to the anatomical joints of human subjects. These robotic orthoses can be divided into two branches namely; user grounded or externally grounded (Stegall et al., 2013). A user grounded robotic orthoses such as Hybrid Assistive Leg (HAL) (Kawamoto & Sankai,

2005; Suzuki, Mito, Kawamoto, Hasegawa, & Sankai, 2007) has no external support and is usually designed for functioning outside rehabilitation clinics. A human subject can perform ADL while wearing these kinds of orthoses. However, they have the weight and power source limitations (Stegall et al., 2013). The externally grounded robotic orthoses such as LOKOMAT (Colombo et al., 2000) have some mechanism to support the weight of orthoses and are typically used in rehabilitation clinics. Most of the recently developed robotic gait training orthoses fall into this category. The design and control of these robotic orthoses has been the prime focus of rehabilitation engineering research community during the past two decades. A review on the design and control of these robotic gait training orthoses has been provided in the author’s previous work (Hussain et al., 2011). Since significant research has been carried out in field of robotic gait rehabilitation orthoses during last three years, this paper presents an updated review of the recent developments in the field. A review on design and control aspects of these robotic orthoses is provided in this paper with a particular emphasis on the research published after the year 2010. The user grounded (Suzuki et al., 2007) and end effector based robotic gait training devices such as Mechanized Gait Trainer (Hesse & Uhlenbrock, 2000) are not included in this review. The single joint robotic orthoses such as Ankle Bot (Roy et al., 2009) and other robotic orthoses intended for rehabilitation of only one joint (Blaya & Herr, 2004; Ferris, Gordon, Sawicki, & Peethambaran, 2006; Grimaldi & Manto; Jamwal, Hussain, & Xie, 2013. DOI: 10.3109/17483107.2013.866986; Jamwal, Xie, Hussain, & Parsons, 2012 (In press) DOI: 10.1109/ TMECH.2012.2219065.) are also not included in the presented review. Similarly, the robotic orthoses utilizing functional electrical stimulation (Stauffer et al., 2009) are also not a part of this review.

2. LOKOMAT LOKOMAT has been developed by researchers at ETH Zurich and works on the basis of a driven gait orthosis (Colombo et al., 2000; Riener et al., 2010). LOKOMAT is one of the only two commercially available robotic gait training orthoses along with AutoAmbulator ("www.deaconess.com,"). However, the details regarding AutoAmbulator have not been extensively reported in literature. The LOKO-

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

MAT has hip and knee sagittal plane rotations powered by Direct current (DC) motors along with a passive foot lifter. LOKOMAT also has a parallelogram mechanism for stabilizing patients’ trunk in vertical plane and avoiding any sideways movement (Colombo et al., 2000). LOKOMAT also provides the options of variable BWS (Frey et al., 2006) and treadmill speed (Von Zitzewitz, Bernhardt, & Riener, 2007) according to the disability level of neurologically impaired patients (Hussain et al., 2011). LOKOMAT has the ability to guide the limbs of patients on desired trajectories by utilizing a trajectory tracking control scheme based on a proportional derivative control scheme (PD) (Colombo et al., 2000). In order to provide customized gait training according the disability level of patients, the developers of LOKOMAT have proposed patient-cooperative gait training strategies based on impedance (Hogan, 1985) and adaptive control algorithms (Banz, Bolliger, M¨uller, Santelli, & Riener, 2009; Duschau-Wicke, Caprez, & Riener, 2010; Jezernik, Colombo, & Morari, 2004; Riener et al., 2005). A path control strategy has also been proposed for providing customized gait training (Alexander Duschau Wicke, 2010; Duschau-Wicke, et al., 2010). These control strategies has been discussed in detail in the review provided by Hussain et al. (2011). Later a bio-cooperative control scheme based on the physiological measurements along with biomechanical measurements from the instrumentation (i.e. force and position sensors) of LOKOMAT has been developed (Koenig, Novak et al.; Koenig, Omlin et al.). Physiological signals such as heart rate (Duschau-Wicke et al., 2010), breathing frequency, skin conductance and skin temperature of healthy and stroke subjects were recorded during LOKOMAT assisted gait training. This bio-cooperative control scheme was used to automatically adapt the virtual training tasks during LOKOMAT assisted gait training so that the patients should experience appropriate level of cognitive load (Koenig, Novak et al.; Koenig, Omlin et al.). The bio-cooperative control schemes was evaluated with stroke patients and provided intended results. A rehabilitation scheme based on Brain-Computer Interface Control (BCI) (King, Wang, Chui, Do, & Nenadic, 2013) of LOKOMAT has also been recently reported in literature (Do, Wang, King, Chun, & Nenadic, 2013). BCI controlled LOKOMAT has been evaluated with one healthy and one paraplegic subject. Electroencephalogram (EEG) data was obtained for both subjects and was used to formulate an EEG prediction model for the online BCI operation (Do, et al.,

703

2013). Preliminary evidence has been recorded to support the fact that the brain-controlled locomotion of SCI subjects is feasible.

3. ALEX Active Leg Exoskeleton (ALEX) has been developed for the gait training of stroke survivors (Banala, Agrawal, Kim, & Scholz, 2010; Banala et al., 2009; Kim et al., 2010). ALEX has been designed by the addition of DC motors at the hip and knee sagittal plane joints of Gravity Balancing Leg Orthosis (GBO). GBO is a passive robotic orhtosis that can compensate the gravitational forces acting on patients’ limbs during gait training (Agrawal et al., 2007; Agrawal & Fattah, 2004; Banala et al., 2006). The actuated hip and knee sagittal plane joints have been controlled by a force field control scheme in order to guide the patients’ foot on predefined trajectories (Banala et al., 2010; Banala et al., 2009). ALEX force field control scheme has shown satisfactory results with stroke patients (Banala et al., 2009). The hip abduction/adduction and trunk rotational degrees of freedom (DOFs) have been held passive with the help of spring mechanisms (Banala et al., 2009). The detailed design and force field control scheme developed for ALEX has also been discussed in detail in the review provided by Hussain et al. (2011). Recently the developers of ALEX have proposed a modified version, ALEX II (Stegall et al., 2013) in order to explore the effect of anterior lunge DOF. Like ALEX, ALEX II also has actuated hip and knee sagittal plane rotations. ALEX II has one additional parallelogram mechanism to provide anterior-posterior trunk motions as compared to ALEX. The similar methods of gravity balancing by using free length springs have also been used in the design of ALEX II like ALEX (Agrawal et al., 2007; Agrawal & Fattah, 2004; Banala et al., 2006). ALEX II has been tested with healthy subjects in order to evaluate the effect of anterior lunge DOF and has provided intended results (Stegall et al., 2013). ALEX II has also been used to evaluate the effects of complementary auditory feedback in lower extremity motor adaptation (Zanotto, Rosati, Spagnol, Stegall, & Agrawal, 2013). Previous studies with ALEX have utilized the concept of introducing motor adaptation by combining kinetic and visual guidance (Banala et al., 2010; Banala et al., 2009). In the recent study, tests with healthy subjects were performed by combining kinetic and visual guidance along with auditory feed-

704

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

back (Zanotto et al., 2013). ALEX II is attached to the limbs of healthy subjects with a pressure sensor mounted shoe. The signals from pressure sensor were used to trigger the rhythmic beats. The results have shown that the combination of kinetic and visual guidance may be as effective as the combination of kinetic and auditory feedback. Adding auditory feedback to kinetic and visual guidance have not lead to significant improvements (Zanotto et al., 2013). ALEX II with only actuated hip sagittal plane joint has recently been used to study the hip and ankle muscle activation patterns (Lenzi, Carrozza, & Agrawal). The evaluations were carried out with healthy individuals and joint angular positions and EMG signals from six muscles of the leg were collected during walking on the treadmill. The subjects walked on the treadmill without wearing ALEX II and baseline kinematics and muscle activation patterns were recorded. After that the subjects walked with ALEX II in zero torque mode and human subjects’ kinematics and muscle activation patterns were recorded. During this mode the robotic exoskeleton did not apply any torque on human subjects’ joints. After zero torque mode the robotic exoskeleton started applying assistive torques gradually to the human subjects’ joints. After this mode the exoskeleton was removed and the kinematics and muscle activation patterns were recorded again to verify any possible alterations of baseline values (Lenzi et al.). Results have shown that the exoskeleton assisted walking reduced the muscle effort compared to free walking (Lenzi et al.).

4. LOPES Lower Extremity Powered Exoskeleton (LOPES) has been developed by the researchers at the University of Twente. LOPES uses the concept of series elastic actuation (SEA) based on Bowden cable based actuation mechanism (Veneman et al., 2006; Veneman et al., 2007). LOPES has two actuated pelvic rotations, hip and knee sagittal plane rotations as well as hip frontal plane rotations (Hussain et al., 2011; Veneman et al., 2006; Veneman et al., 2007). Two gait training modes namely; patient-in charge mode and robot-in charge mode have been utilized by LOPES. During patient-in charge or zero torque mode, the robotic exoskeleton was completely passive and the human subjects’ were voluntarily driving the exoskeleton. Gait kinematics and muscle activations patterns have been recorded previously to evaluate the performance of LOPES during

patient-in charge mode (Van Asseldonk et al., 2008; Veneman et al., 2007). Previously LOPES has not been evaluated in robot-in charge mode (Van Asseldonk, et al., 2008; Veneman et al., 2007). Recently, the developers of LOPES have implemented a Virtual Model Controller (VMC) for the robot-in charge mode and evaluated it for the foot clearance task with healthy and chronic stroke subjects (Koopman, Van Asseldonk, & Van Der Kooij). The VMC provides robotic support at the ankle and aides in increasing the foot clearance during treadmill walking task. Physical interactions with the patient were defined in the VMC that would assist the gait subtasks (i.e. step height). These interactions were then modelled as springs and dampers that can be switched at appropriate intervals during gait cycle (Koopman et al.). An errordriven adaptation algorithm developed by Emken et al. (Emken, Harkema, Beres-Jones, Ferreira, & Reinkensmeyer, 2008) has been utilized to adapt the level of support provided by LOPES. The adaptation algorithm adjusts the virtual spring stiffness at each phase of gait cycle based on the error recorded during the previous step (Koopman et al.). This adaptation algorithm has been shown effective and has shaped the support level to specific needs of stroke survivors. The VMC has also helped in selectively and gradually influencing the step height (Koopman et al.).

5. ICRO Intrinsically Compliant Robotic Orthosis (ICRO) has been developed at the University of Auckland (Hussain et al., 2012). ICRO has been developed based on the concept of light weight robotic orthosis that can provide safe human-robot interaction. Hip and knee sagittal plane rotations of ICRO have been powered by intrinsically compliant Pneumatic Muscle Actuators (PMA). PMA has a working principle similar to skeletal muscles and provides actuation to single joint in the form of pairs (i.e. antagonistic actuation). Trunk of the ICRO has two passive DOFs namely vertical and lateral translations. Hip joint also has abduction/adduction motions and has been held passive. A passive foot lifter has also been provided to ensure sufficient foot clearance. All the segments of the ICRO has been made telescopic (Hussain et al., 2012). A spring was used to compensate the weight of ICRO. A trajectory tracking control scheme based on proportional control has been developed and evaluated with healthy individuals to evaluate the performance

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

of ICRO (Hussain et al., 2012). The proportional control provided the intended results and ICRO was able to guide the subject’s limbs on reference trajectories (Hussain et al., 2012). As ICRO has been actuated by PMA, which are highly nonlinear and time varying, control of ICRO has been a difficult task. In order to encounter the nonlinear and time varying behaviour of PMA, trajectory tracking control schemes based on robust sliding mode control laws has also been developed (Hussain, Xie, & Jamwal, 2013; Hussain, Xie, & Jamwal, 2013b). The robust trajectory tracking control schemes have been evaluated with healthy subjects (Hussain et al., 2013; Hussain et al., 2013-b). The evaluations have shown that these robust control laws can guide the subjects’ limbs on predefined trajectories despite the nonlinear and time varying behaviour of PMA (Hussain et al., 2013; Hussain et al., 2013-b). In order to provide customized and assist-as-needed gait training, an adaptive impedance control scheme has also been developed for ICRO (Hussain, Xie, & Jamwal, 2013-a). The adaptive impedance control scheme utilizes sliding mode control law as the basic position controller and is capable of adapting the robotic assistance according to the disability level and stage of rehabilitation of neurologically impaired subjects (Hussain et al., 2013-a). The adaptive impedance control scheme measures the human subject voluntary participation from the force sensors data (Banz et al., 2009) which were placed in series with the PMA. Based on the force sensor data and the inverse dynamics model of human lower limbs and ICRO, the active human joint torque component was estimated. This active human joint torque component was then used to adapt the assistance provided by ICRO (Hussain et al., 2013-a). The adaptive impedance control scheme has been evaluated with neurologically intact subjects and the assistance provided by ICRO decreased when the human subjects increased their voluntary participation (Hussain et al., 2013-a).

6. Discussion Robot assisted gait training is an emerging physical rehabilitation practice. A considerable volume of research has been performed on the design and control of robotic gait rehabilitation orthoses during the past two decades (Hussain et al., 2011). These robotic orthoses work in close proximity with human subjects’ anatomical joints and have the potential of providing customized gait training according to the disability level

705

and stage of rehabilitation of neurologically impaired patients (Riener et al., 2005). An updated review on the design and control aspects of robotic gait rehabilitation orthoses is presented. Recent developments in the design of robotic gait rehabilitation orthoses have resulted in providing additional DOFs to the robotic orthoses (Stegall et al., 2013). The number of DOFs has been an important factor in the design of robotic gait rehabilitation orthoses (Hussain, Xie, & Jamwal, 2013) as the biomechanics and rehabilitation engineering community has a limited understanding of the kinematics and kinetics of human gait. Based on the current understanding of kinematics of human gait, the rehabilitation robotics community has attempted to choose an optimal set of DOFs. The initial prototypes such as LOKOMAT have only included sagittal plane motions in the design of robotic orthoses (Colombo et al., 2000). Later DOFs in frontal and transverse planes have also been included (Banala et al., 2009; Hussain et al., 2012; Stegall, et al., 2013; Van Asseldonk et al., 2008). However, providing additional DOFs have proven to be a challenge as it increases the complexity of mechanism and further leads to safety and control problems. The number of actuated and passive DOFs of the robotic gait training orthoses has also remained an unanswered research question during the past decade. The rehabilitation robotics community has been trying to choose an optimal set of actuated DOFs as actuating all the DOFs introduces mechanism design and control complexities. The initial prototypes of robotic orthoses such as LOKOMAT have only actuated hip and knee sagittal plane rotations (Colombo et al., 2000). The latest version of LOKOMAT still has the same actuated DOFs. All the other DOFs are passive. Similarly, the latest prototypes of ALEX II (Stegall et al., 2013) and ICRO (Hussain et al., 2012) also have actuated hip and knee sagittal plane rotations. Although the force field control algorithm of ALEX can guide the patients’ foot on predefined trajectories (Banala et al., 2009). Later the developers of LOPES have included more actuated DOFS in the frontal and transverse planes (Veneman et al., 2007). They have evaluated walking in a passive LOPES (i.e. non powered) and compared it with free walking on a treadmill (Van Asseldonk et al., 2008). The results of both those modes were compared and have shown little difference between LOPES walking and free treadmill walking (Van Asseldonk et al., 2008). However, findings of walking with powered (i.e. actuated) LOPES have not been extensively reported in literature.

706

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

Neurologic injuries such as stroke often result in drop foot issues and ankle joint impairments. Mechanism design and actuation of ankle joint has been a critical issue in the above mentioned robotic gait training orthoses. Robotic orthoses design for ankle joint is a difficult mechanical task as ankle is a complex anatomical joint. Several robotic orthoses has been developed for ankle joint rehabilitation (Blaya & Herr, 2004; Ferris et al., 2006; Roy et al., 2009). Unfortunately, these robotic ankle orthoses have not been incorporated with the robotic gait training orthoses. Robotic gait training orthoses such as LOKOMAT (Colombo et al., 2000) and ICRO (Hussain et al., 2012) incorporate a passive foot lifter to provide sufficient foot clearance in sagittal plane during the treadmill walking task. ALEX has not an actuated ankle joint but the force field control algorithm guides the patients’ foot on predefined trajectories in sagittal plane (Banala et al., 2009). LOPES has no actuated or passive ankle joint (Veneman et al., 2007). This presents a limitation of current robotic gait training orthoses and much work needs to be done in either designing new anatomical ankle robotic orthoses or adding currently available ankle orthoses to the robotic gait training orthoses. The choice of actuators for powering the robotic orthoses has also been an active research question. The initial prototype of robotic orthoses such as LOKOMAT (Colombo et al., 2000) and ALEX (Banala et al., 2009) incorporates electromechanical actuators such as servo or DC motors. These actuators provide good trajectory tracking control, however, they have high weight and high end point impedance and may not be suitable for providing safe human-robot interaction (Hussain et al., 2012). The use of heavy electromechanical systems increase the weight of robotic orthoses which may not be suitable for implementing advance assist-as-needed gait training strategies (Hussain et al., 2012). Later, robotic orthoses based on the concept of compliant actuation such as LOPES (Veneman et al., 2007) and ICRO (Hussain et al., 2012) have been developed. These compliant robotic orthoses are built on the concept of light weight actuation mechanisms and are capable of providing safe human-robot interaction. The alignment of robotic gait training orthoses joints with human subjects’ anatomical joints is another critical research issue (Colombo et al., 2000). The robotic orthoses joints must be correctly aligned with the anatomical joints for proper functioning. If the alignment is not done properly, the patient may feel discomfort and the robot assisted gait therapy may not

yield intended results. This alignment is a difficult task and may take 20–30 minutes for each patient before the start of gait training. This phenomenon has also been reported during the initial clinical trials of LOKOMAT (Colombo et al., 2000). Even if the alignment is done properly at the beginning of each training session, the robotic orhtosis may slip during the session due to the weight of orthosis and relative movement with skin of human subject (Colombo et al., 2000; Stienen, Hekman, van der Helm, & van der Kooij, 2009). Moreover, the same alignment may not be restored for a patient during different gait training sessions. Recently, attempts have been made to reduce the weight of robotic gait training orthoses (Hussain et al., 2012; Veneman et al., 2007); however, the issue of alignment is still unresolved. In the recent years significant work has been performed in designing the robotic orthoses for upper limb rehabilitation that can closely align with human subjects’ anatomical joints (Cempini, De Rossi, Lenzi, Vitiello, & Carrozza, 2013; Jarrass´e & Morel, 2012; Schiele & Van Der Helm, 2006; Stienen et al., 2009). However, no significant work has been reported in literature regarding the design of robotic gait training orthoses that can closely align with anatomical joints. The control of these robotic gait training orthoses for providing customized rehabilitation is also a continuously evolving research area. Most of the commonly used control strategies have been previously discussed in (Hussain et al., 2011). Recently, bio-cooperative control schemes based on physiological and biomechanical measurements have been developed and evaluated for LOKOMAT (Koenig, Novak et al.; Koenig, Omlin et al.). A control scheme for LOKOMAT based on BCI has also been developed and evaluated with health and neurologically impaired subjects and has proven to be feasible for gait rehabilitation (Do et al., 2013). Similarly, a control scheme has also been formulated for ALEX II by combining kinetic and visual guidance along with auditory feedback and has been evaluated with healthy subjects (Zanotto et al., 2013). Control of robotic gait training orthoses powered by compliant or SEA is an important and difficult task (Hussain et al., 2012; Veneman et al., 2007). Considerable research has been reported in the field of control of compliant robotic gait training orthoses in recent years. A VMC along with an error-driven adaptation algorithm has been developed and evaluated with neurologically impaired patients for the robot-in charge mode of LOPES (Koopman et al.). These trials have provided satisfactory results for the robot-in-charge mode of LOPES (Koopman et al.). A robust control

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

scheme has been developed and evaluated with healthy subjects for ICRO assisted gait training. This robust control scheme has provided satisfactory trajectory tracking results despite the nonlinear and time varying behaviour of PMA (Hussain et al., 2013-b). An adaptive impedance control scheme has also been developed and evaluated with healthy subjects for ICRO. The adaptive impedance control scheme has also provided intended results (Hussain et al., 2013-a). These recent developments have provided an encouraging evidence for the use of intrinsically compliant actuators in the field of robot assisted gait rehabilitation. In conclusion, the field of robot assisted gait training has made significant progress during the past two decades. Several robotic orthoses have been designed and novel control strategies have been implemented that can provide customized gait rehabilitation. These robotic orthoses and control strategies have shown promising results from preliminary clinical trials. Much research needs to be done in order to develop better robotic mechanisms and actuation methods that can provide naturalistic gait patterns and safe human-robot interaction during gait training. Similarly, the control strategies for these robotic orthoses need to be improved so that customized gait rehabilitation can be provided. More extensive clinical trials also need to be performed to establish the efficacy of these robotic orthoses and control strategies.

Acknowledgments The author would like to acknowledge the seed research grant provided by Nazarbayev University for the project titled “An Intrinsically Compliant Robotic Orthosis for Customized Gait Rehabilitation” under Grant KF-14/07.

Competing interests The author declares no competing interests with any person or organization.

References Agrawal, S. K., Banala, S. K., Fattah, A., Sangwan, V., Krishnamoorthy, V., Scholz, J. P. et al. (2007). Assessment of Motion of a Swing Leg and Gait Rehabilitation With a Gravity Balancing Exoskeleton. IEEE Transactions on Neural systems And Rehabilitation Engineering, 15(3), 410-420.

707

Agrawal, S. K., & Fattah, A. (2004). Theory and design of an orthotic device for full or partial gravity-balancing of a human leg during motion. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 12(2), 157-165. Alexander Duschau Wicke, J. v. Z., Andrea Caprez, Lars Lunenburger, & Robert Riener. (2010). Path Control: A Method for Patient-Cooperative Robot Aided Gait Rehabilitation.IEEE Transactions on Neural Systems And Rehabilitation Engineering, 18(1), 38-48. Banala, S. K., Agrawal, S.K., Fattah, A., Krishnamoorthy, V., Hsu, W. L., Scholz, J. et al. (2006). Gravity-balancing leg orthosis and its performance evaluation. IEEE Transactions on Robotics, 22(6), 1228-1239. Banala, S. K., Agrawal, S. K., Kim, S. H., & Scholz, J. P. (2010). Novel gait adaptation and neuromotor training results using an active leg exoskeleton. IEEE/ASME Transactions on Mechatronics, 15(2), 216-225. Banala, S. K., Kim, S. H., Agrawal, S. K., & Scholz, J. P. (2009). Robot assisted gait training with active leg exoskeleton (ALEX). IEEE Transactions on Neural Systems and Rehabilitation Engineering, 17(1), 2-8. Banz, R., Bolliger, M., M¨uller, S., Santelli, C., & Riener, R. (2009). A method of estimating the degree of active participation during stepping in a driven gait orthosis based on actuator force profile matching. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 17(1), 15-22. Barbeau, H. (1999). Walking after spinal cord injury: Evaluation, treatment, and functional recovery. Archives of Physical Medicine and Rehabilitation, 80, 225-235. Barbeau, H., & Fung, J. (2001). The role of rehabilitation in the recovery of walking in the neurological population. Current Opinion in Neurology, 14(6), 735-740. Blaya, J. A., & Herr, H. (2004). Adaptive Control of a VariableImpedance Ankle-Foot Orthosis to Assist Drop-Foot Gait. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 12 (1), 24-31. Cempini, M., De Rossi, S. M. M., Lenzi, T., Vitiello, N., & Carrozza, M. C. (2013). Self-alignment mechanisms for assistive wearable robots: A kinetostatic compatibility method. IEEE Transactions on Robotics, 29(1), 236-250. Colombo, G., Joerg, M., Schreier, R., & Dietz, V. (2000). Treadmill training of paraplegic patients using a robotic orthosis. Journal of Rehabilitation Research and Development, 37(6), 693-700. Cramer, S. C., & Riley, J. D. (2008). Neuroplasticity and brain repair after stroke. Current Opinion in Neurology, 21(1), 76-82. Da Cunha, I. T. Jr., Lim, P. A., Qureshy, H., Henson, H., Monga, T., & Protas, E. J. (2002). Gait outcomes after acute stroke rehabilitation with supported treadmill ambulation training: A randomized controlled pilot study. Archives of Physical Medicine and Rehabilitation, 83(9), 1258-1265. Daisuke Aoyagi, W. E. I., Susan, J., Harkema, David, J., Reinkensmeyer, James, E., Bobrow, (2007). A Robot and Control Algorithm That Can Synchronously Assist in Naturalistic Motion During Body-Weight-Supported Gait Training Following Neurologic Injury. IEEE Transactions on Neural Systems And Rehabilitation Engineering, 15(3), 387-400. Do, A. H., Wang, P. T., King, C. E., Chun, S. N., & Nenadic, Z. (2013). Brain-computer interface controlled robotic gait orthosis. Journal of NeuroEngineering and Rehabilitation, 10, 111. Duschau-Wicke, A., Caprez, A., & Riener, R. (2010). Patientcooperative control increases active participation of individuals

708

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses

with SCI during robot-aided gait training. Journal of NeuroEngineering and Rehabilitation, 7(1). Emken, J. L., Harkema, S. J., Beres-Jones, J. A., Ferreira, C. K., & Reinkensmeyer, D. J. (2008). Feasibility of manual teach-andreplay and continuous impedance shaping for robotic locomotor training following spinal cord injury. IEEE Transactions on Biomedical Engineering, 55(1), 322-334. Ferris, D. P., Gordon, K. E., Sawicki, G. S., & Peethambaran, A. (2006). An improved powered ankle-foot orthosis using proportional myoelectric control. Gait and Posture, 23(4), 425428. French, B., Thomas, L. H., Leathley, M. J., Sutton, C. J., McAdam, J., Forster, A. et al. (2007). Repetitive task training for improving functional ability after stroke. Cochrane Database of Systematic Reviews(4). Frey, M., Colombo, G., Vaglio, M., Bucher, R., J¨org, M., & Riener, R. (2006). A novel mechatronic body weight support system. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 14(3), 311-321. Grimaldi, G., & Manto, M. Functional impacts of exoskeleton-based rehabilitation in chronic stroke : Multi-joint versus single-joint robotic training. Journal of NeuroEngineering and Rehabilitation, 113. Hassid, E., Rose, D., Commisarow, J., Guttry, M., & Dobkin, B. H. (1997). Improved gait symmetry in hemiparetic stroke patients induced during body weight-supported treadmill stepping. Journal of Neurologic Rehabilitation, 11(1), 21-26. Hesse, S., Bertelt, C., Jahnke, M. T., Schaffrin, A., Baake, P., Malezic, M. et al. (1995). Treadmill training with partial body weight support compared with physiotherapy in nonambulatory hemiparetic patients. Stroke, 26(6), 976-981. Hesse, S., Konrad, M., & Uhlenbrock, D. (1999). Treadmill walking with partial body weight support versus floor walking in hemiparetic subjects. Archives of Physical Medicine and Rehabilitation, 80(4), 421-427. Hesse, S., & Uhlenbrock, D. (2000). A mechanized gait trainer for restoration of gait. Journal of Rehabilitation Research and Development, 37(6), 701-708. Hogan, N. (1985). Impedance Control: An approach to manipulation. J Dynamic Syst Meas Control, 107, 1-23. Holder, N. L., Clark, H. A., DiBlasio, J. M., Hughes, C. L., Scherpf, J. W., & Harding, L. et al. (1999). Cause, prevalence, and response to occupational musculoskeletal injuries reported by physical therapists and physical therapist assistants. Physical Therapy, 79(7), 642-652. Hussain, S., Xie, S. Q., & Jamwal, P. K. (2013). Control of a Robotic Orhtosis for Gait Rehabilitation. Robotics and Autonomous Systems, 61, 911-919. Hussain, S., Xie, S. Q., & Jamwal, P. K. (2013-a). Adaptive Impedance Control of a Robotic Orthosis for Gait Rehabilitation. IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 43(3), 1025-1034. Hussain, S., Xie, S. Q., & Jamwal, P. K. (2013). Effect of Cadence Regulation on Muscle Activation Patterns during Robot Assisted Gait: A Dynamic Simulation Study. IEEE Journal of Biomedical and Health Informatics, 17(2), 442-451. Hussain, S., Xie, S. Q., & Jamwal, P. K. (2013-b). Robust Nonlinear Control of an Intrinsically Compliant Robotic Gait Training Orthosis. IEEE Transactions on Systems, Man, and Cybernetics—Part A: Systems and Humans, 43(3), 655665.

Hussain, S., Xie, S. Q., Jamwal, P. K., & Parsons, J. (2012). An Intrinsically Compliant Robotic Orthosis for Treadmill Training. Medical Engineering and Physics, 34, 1448-1453. Hussain, S., Xie, S. Q., & Liu, G. (2011). Robot assisted treadmill training: Mechanisms and training strategies. Medical Engineering & Physics, 33(5), 527-533. Jamwal, P. K., Hussain, S., & Xie, S. Q. (2013. DOI: 10.3109/17483107.2013.866986). Review on design and control aspects of ankle rehabilitation robots. Disability and Rehabilitation: Assistive Technology. Jamwal, P. K., Xie, S. Q., Hussain, S., & Parsons, J. (2012 (In press) DOI: 10.1109/TMECH.2012.2219065.). An Adaptive Wearable Parallel Robot for Ankle Injury Treatments. IEEE/ASME Transactions on Mechatronics. Jarrass´e, N., & Morel, G. (2012). Connecting a human limb to an exoskeleton. IEEE Transactions on Robotics, 28(3), 697-709. Jezernik, S., Colombo, G., & Morari, M. (2004). Automatic gaitpattern adaptation algorithms for rehabilitation with a 4-DOF robotic orthosis. IEEE Transactions on Robotics and Automation, 20(3), 574-582. Kawamoto, H., & Sankai, Y. (2005). Power assist method based on Phase Sequence and muscle force condition for HAL. Advanced Robotics, 19(7), 717-734. Kim, S. H., Banala, S. K., Brackbill, E. A., Agrawal, S. K., Krishnamoorthy, V., & Scholz, J.P. (2010). Robot-assisted modifications of gait in healthy individuals. Experimental Brain Research, 202(4), 809-824. King, C. E., Wang, P. T., Chui, L. A., Do, A. H., & Nenadic, Z. (2013). Operation of a brain-computer interface walking simulator for individuals with spinal cord injury. Journal of NeuroEngineering and Rehabilitation, 10, 77. Koenig, A., Novak, D., Omlin, X., Pulfer, M., Perreault, E., & Zimmerli, L. et al. Real-time closed-loop control of cognitive load in neurological patients during robot-assisted gait training. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 19(4) 453-464. Koenig, A., Omlin, X., Bergmann, J., Zimmerli, L., Bolliger, M., Muller, F., et al. Controlling patient participation during robot-assisted gait training. Journal of NeuroEngineering and Rehabilitation, 8(1). Koopman, B., Van Asseldonk, E. H. F., & Van Der Kooij, H. Selective control of gait subtasks in robotic gait training: Foot clearance support in stroke survivors with a powered exoskeleton. Journal of NeuroEngineering and Rehabilitation, 10(1). Kwakkel, G., Wagenaar, R. C., Twisk, J. W. R., Lankhorst, G. J., & Koetsier, J. C. (1999). Intensity of leg and arm training after primary middle-cerebral-artery stroke: A randomised trial. Lancet, 354(9174), 191-196. Laufer, Y., Dickstein, R., Chefez, Y., & Marcovitz, E. (2001). The effect of treadmill training on the ambulation of stroke survivors in the early stages of rehabilitation: A randomized study. Journal of Rehabilitation Research and Development, 38(1), 69-78. Lenzi, T., Carrozza, M. C., & Agrawal, S. K., Powered hip exoskeletons can reduce the user’s hip and ankle muscle activations during walking. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 21(6), 938-948. McCain, K. J., Pollo, F. E., Baum, B. S., Coleman, S. C., Baker, S., & Smith, P. S.(2008).Locomotor Treadmill Training With Partial Body-Weight Support Before Overground Gait in Adults With Acute Stroke: A Pilot Study. Archives of Physical Medicine and Rehabilitation, 89(4), 684-691.

S. Hussain / State-of-the-art robotic gait rehabilitation orthoses Moreno, J. C., Barroso, F., Farina, D., Gizzi, L., Santos, C., Molinari, M., et al. Effects of robotic guidance on the coordination of locomotion. Journal of NeuroEngineering and Rehabilitation, 10 (1). Mulroy, S. J., Klassen, T., Gronley, J. K., Eberly, V. J., Brown, D. A., & Sullivan, K. J. (2010). Gait parameters associated with responsiveness to treadmill training with body-weight support after stroke: An exploratory study. Physical Therapy, 90(2), 209223. Murray, M. P., Spurr, G. B., & Sepic, S. B. (1985). Treadmill vs. floor walking: Kinematics, electromyogram, and heart rate. Journal of Applied Physiology, 59(1), 87-91. Patterson, S. L., Rodgers, M. M., Macko, R. F., & Forrester, L. W. (2008). Effect of treadmill exercise training on spatial and temporal gait parameters in subjects with chronic stroke: A preliminary report. Journal of Rehabilitation Research and Development, 45(2), 221-228. Pennycott, A., Wyss, D., Vallery, H., Klamroth-Marganska, V., & Riener, R. Towards more effective robotic gait training for stroke rehabilitation: A review. Journal of NeuroEngineering and Rehabilitation, 9(1). Riener, R., L¨unenburger, L., Maier, I. C., Colombo, G., & Dietz, V. (2010). Locomotor training in subjects with sensori-motor deficits: An overview of the robotic gait orthosis Lokomat. Journal of Healthcare Engineering, 1(2), 197-216. Riener, R., L¨unenburger, L., Jezernik, S., Anderschitz, M., Colombo, G., & Dietz, V. (2005). Patient-cooperative strategies for robot-aided treadmill training: First experimental results. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 13(3), 380-394. Roger, V. L., Lloyd-Jones, D. M., & Berry, F. J. D. (2011). Heart disease and stroke statistics-2011 update: A report from the American Heart Association. Circulation, 123(4), e18-e19 Roy, A., Krebs, H. I., Williams, D. J., Bever, C. T., Forrester, L. W., & Macko, R. M. et al. (2009). Robot-aided neurorehabilitation: A novel robot for ankle rehabilitation. IEEE Transactions on Robotics, 25(3), 569-582. Schiele, A., & Van Der Helm, F. C. T. (2006). Kinematic design to improve ergonomics in human machine interaction. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 14(4), 456-469. Stauffer, Y., Allemand, Y., Bouri, M., Fournier, J., Clavel, R., Metrailler, P. et al. (2009). The WalkTrainer - A new generation of walking reeducation device combining orthoses and muscle stimulation. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 17(1), 38-45. Stegall, P., Winfree, K., Zanotto, D., & Agrawal, S. K. (2013). Rehabilitation Exoskeleton Design: Exploring the Effect of the

709

Anterior Lunge Degree of Freedom. IEEE Transactions on Robotics, 29(4), 838-846. Stienen, A. H. A., Hekman, E. E. G., van der Helm, F. C., T., & van der Kooij, H. (2009). Self-aligning exoskeleton axes through decoupling of joint rotations and translations. IEEE Transactions on Robotics, 25(3), 628-633. Suzuki, K., Mito, G., Kawamoto, H., Hasegawa, Y., & Sankai, Y. (2007).Intention-based walking support for paraplegia patients with Robot Suit HAL. Advanced Robotics, 21(12), 14411469. Teixeira da Cunha Filho, I., Lim, P. A. C., Qureshy, H., Henson, H., Monga, T., & Protas, E. J. (2001). A comparison of regular rehabilitation and regular rehabilitation with supported treadmill ambulation training for acute stroke patients. Journal of Rehabilitation Research and Development, 38(2), 245-255. Van Asseldonk, E. H. F., Veneman, J. F., Ekkelenkamp, R., Buurke, J. H., Van Der Helm, F. C. T., & Van Der Kooij, H. (2008). The effects on kinematics and muscle activity of walking in a robotic gait trainer during zero-force control. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 16(4), 360370. Van Peppen, R. P. S., Kwakkel, G., Wood-Dauphinee, S., Hendriks, H. J. M., Van der Wees, P. J., & Dekker, J. (2004). The impact of physical therapy on functional outcomes after stroke: What’s the evidence? Clinical Rehabilitation, 18(8), 833-862. Veneman, J. F., Ekkelenkamp, R., Kruidhof, R., Van Der Helm, F. C. T., & Van Der Kooij, H. (2006). A series elastic- and bowden-cable-based actuation system for use as torque actuator in exoskeleton-type robots. International Journal of Robotics Research, 25(3), 261-281. Veneman, J. F., Kruidhof, R., Hekman, E. E. G., Ekkelenkamp, R., Van Asseldonk, E. H. F., & Van Der Kooij, H. (2007). Design and evaluation of the LOPES exoskeleton robot for interactive gait rehabilitation. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 15(3), 379-386. Visintin, M., Barbeau, H., Korner-Bitensky, N., & Mayo, N. E. (1998). A new approach to retrain gait in stroke patients through body weight support and treadmill stimulation. Stroke, 29(6), 11221128. Von Zitzewitz, J., Bernhardt, M., & Riener, R. (2007). A novel method for automatic treadmill speed adaptation. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 15(3), 401-409. www.deaconess.com. Zanotto, D., Rosati, G., Spagnol, S., Stegall, P., & Agrawal, S. K. (2013). Effects of Complementary Auditory Feedback in RobotAssisted Lower Extremity Motor Adaptation. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 21(5), 775786.