Trends in robot-assisted and virtual reality-assisted neuromuscular therapy: a systematic review of health-related multiplayer games.

BACKGROUND: Multiplayer games have emerged as a promising approach to increase the motivation of patients involved in rehabilitation therapy. In this systematic review, we evaluated recent publications in health-related multiplayer games that involved patients with cognitive and/or motor impairments. The aim was to investigate the effect of multiplayer gaming on game experience and game performance in healthy and non-healthy populations in comparison to individual game play. We further discuss the publications within the context of the theory of flow and the challenge point framework. METHODS: A systematic search was conducted through EMBASE, Medline, PubMed, Cochrane, CINAHL and PsycINFO. The search was complemented by recent publications in robot-assisted multiplayer neurorehabilitation. The search was restricted to robot-assisted or virtual reality-based training. RESULTS: Thirteen articles met the inclusion criteria. Multiplayer modes used in health-related multiplayer games were: competitive, collaborative and co-active multiplayer modes. Multiplayer modes positively affected game experience in nine studies and game performance in six studies. Two articles reported increased game performance in single-player mode when compared to multiplayer mode. CONCLUSIONS: The multiplayer modes of training reviewed improved game experience and game performance compared to single-player modes. However, the methods reviewed were quite heterogeneous and not exhaustive. One important take-away is that adaptation of the game conditions can individualize the difficulty of a game to a player's skill level in competitive multiplayer games. Robotic assistance and virtual reality can enhance individualization by, for example, adapting the haptic conditions, e.g. by increasing haptic support or by providing haptic resistance. The flow theory and the challenge point framework support these results and are used in this review to frame the idea of adapting players' game conditions.

Effects of robotically modulating kinematic variability on motor skill learning and motivation.

It is unclear how the variability of kinematic errors experienced during motor training affects skill retention and motivation. We used force fields produced by a haptic robot to modulate the kinematic errors of 30 healthy adults during a period of practice in a virtual simulation of golf putting. On day 1, participants became relatively skilled at putting to a near and far target by first practicing without force fields. On day 2, they warmed up at the task without force fields, then practiced with force fields that either reduced or augmented their kinematic errors and were finally assessed without the force fields active. On day 3, they returned for a long-term assessment, again without force fields. A control group practiced without force fields. We quantified motor skill as the variability in impact velocity at which participants putted the ball. We quantified motivation using a self-reported, standardized scale. Only individuals who were initially less skilled benefited from training; for these people, practicing with reduced kinematic variability improved skill more than practicing in the control condition. This reduced kinematic variability also improved self-reports of competence and satisfaction. Practice with increased kinematic variability worsened these self-reports as well as enjoyment. These negative motivational effects persisted on day 3 in a way that was uncorrelated with actual skill. In summary, robotically reducing kinematic errors in a golf putting training session improved putting skill more for less skilled putters. Robotically increasing kinematic errors had no performance effect, but decreased motivation in a persistent way.

Time flies when you are in a groove: using entrainment to mechanical resonance to teach a desired movement distorts the perception of the movement's timing.

The human motor system quickly entrains rhythmic limb movement to the resonant frequency of mechanical systems with which it interacts, suggesting that entrainment to an appropriately designed training device might be a convenient way to teach desired movements. We tested this possibility by asking healthy subjects (N = 30) to learn to move with a desired movement timing using a simple resonating arm training device: a lever attached to a manual wheelchair. The subjects tried to learn to roll the lever-driven wheelchair back and forth in place at a target frequency initially presented using a series of auditory beeps. One-third of the subjects trained without resonance and with no further feedback about rolling frequency; their performance did not improve. Another group trained with continual visual feedback of frequency error but no resonance; they quickly learned to roll the chair at the target frequency, as evidenced at both short-term and long-term (1 day later) retention tests. A third group trained with elastic bands attached to the lever that caused the system to resonate at the target frequency, providing a timing template. While these participants quickly entrained to the target frequency during training, they did not accurately reproduce this frequency when the system was no longer resonant, moving too slowly with the same systematic error at both the short-term and long-term retention tests. They also did not exhibit a timing aftereffect on the initial movements made when they transitioned from a resonant to non-resonant system or vice versa. This suggests they did not realize they were performing the task with a temporal error. Entrainment to mechanical resonance conveys usable information about movement timing, but seems to cause that movement timing to be perceived as slower than it actually is, as if a putative internal clock speeds up, which is a factor to consider in designing machine-assisted motor training.

Barriers and Facilitators to the Use of Wearable Robots as Assistive Devices: Qualitative Study With Older Adults and Physiotherapists.

Background: Light wearable robots have the potential to assist older adults with mobility impairments in daily life by compensating for age-related decline in lower extremity strength. Physiotherapists may be the first point of contact for older adults with these devices. Objective: The aims of this study were to explore views of older adults and physiotherapists on wearable robots as assistive devices for daily living and to identify the barriers and facilitators to their use. Methods: Six older adults (aged 72-88 years) tested a wearable robot (Myosuit) and participated in semistructured interviews. A focus group with 6 physiotherapists who had a minimum of 5 years of professional experience and specialized in geriatrics was conducted. Data were analyzed using thematic qualitative text analysis. Results: Older adults perceived benefits and had positive use experiences, yet many saw no need to use the technology for themselves. Main barriers and facilitators to its use were the perception of usefulness, attitudes toward technology, ease of use, and environmental factors such as the support received. Physiotherapists named costs, reimbursement schemes, and complexity of the technology as limiting factors. Conclusions: A light wearable robot-the Myosuit-was found to be acceptable to study participants as an assistive device. Although characteristics of the technology are important, the use and acceptance by older adults heavily depend on perceived usefulness and need.

Guided Exploration Leads to Faster Familiarization with a Wearable Robot: First Results of an Innovative Protocol.

Wearable robots show promise in addressing physical and functional deficits in individuals with mobility impairments. However, the process of learning to use these devices can take a long time. In this study, we propose a novel protocol to support the familiarization process with a wearable robot (the Myosuit) and achieve faster walking speeds. The protocol involves applying an anterior pulling force while participants perform a series of 10-meter Walking Tests (10mWT) with or without the Myosuit under various experimental conditions. We hypothesized that guiding the exploration of novel walking patterns can help the users learn to exploit the Myosuit's assistance faster by leading to larger step lengths and ultimately higher walking speeds. In this paper, we present the preliminary results of the protocol with seven participants with lower-limb mobility impairments. Participants who were assisted by the Myosuit showed a continuous increase in walking speed over the course of the pulling part of the experiment with a maximum increase of 41.3% (10.4%) when compared to the baseline 10mWT. Following the removal of the pulling force, these participants continued to show an increased walking speed while being supported by the Myosuit. This higher walking speed was primarily due to a significant increase in step length of 24% (16.6%) and cadence of 11% (8.9%). The results of this study may help the development of familiarization techniques for wearable robots.

An algorithm to reduce human-robot interface compliance errors in posture estimation in wearable robots.

Assistive forces transmitted from wearable robots to the robot's users are often defined by controllers that rely on the accurate estimation of the human posture. The compliant nature of the human-robot interface can negatively affect the robot's ability to estimate the posture. In this article, we present a novel algorithm that uses machine learning to correct these errors in posture estimation. For that, we recorded motion capture data and robot performance data from a group of participants (n = 8; 4 females) who walked on a treadmill while wearing a wearable robot, the Myosuit. Participants walked on level ground at various gait speeds and levels of support from the Myosuit. We used optical motion capture data to measure the relative displacement between the person and the Myosuit. We then combined this data with data derived from the robot to train a model, using a grading boosting algorithm (XGBoost), that corrected for the mechanical compliance errors in posture estimation. For the Myosuit controller, we were particularly interested in the angle of the thigh segment. Using our algorithm, the estimated thigh segment's angle RMS error was reduced from 6.3° (2.3°) to 2.5° (1.0°), mean (standard deviation). The average maximum error was reduced from 13.1° (4.9°) to 5.9° (2.1°). These improvements in posture estimation were observed for all of the considered assistance force levels and walking speeds. This suggests that ML-based algorithms provide a promising opportunity to be used in combination with wearable-robot sensors for an accurate user posture estimation.

Age-Dependent Asymmetry of Wrist Position Sense Is Not Influenced by Stochastic Tactile Stimulation.

Stochastic stimulation has been shown to improve movement, balance, the sense of touch, and may also improve position sense. This stimulation can be non-invasive and may be a simple technology to enhance proprioception. In this study, we investigated whether sub-threshold stochastic tactile stimulation of mechanoreceptors reduces age-related errors in wrist position estimation. Fifteen young (24.5±1.5y) and 23 elderly (71.7±7.3y) unimpaired, right-handed adults completed a wrist position gauge-matching experiment. In each trial, the participant's concealed wrist was moved to a target position between 10 and 30° of wrist flexion or extension by a robotic manipulandum. The participant then estimated the wrist's position on a virtual gauge. During half of the trials, sub-threshold stochastic tactile stimulation was applied to the wrist muscle tendon areas. Stochastic stimulation did not significantly influence wrist position sense. In the elderly group, estimation errors decreased non-significantly when stimulation was applied compared to the trials without stimulation [mean constant error reduction Δ µ ( θ c o n o f ) = 0 . 8 ° in flexion and Δ µ ( θ c o n o e ) = 0 . 7 ° in extension direction, p = 0.95]. This effect was less pronounced in the young group [ Δ µ ( θ c o n y ) = 0 . 2 ° in flexion and in extension direction, p = 0.99]. These improvements did not yield a relevant effect size (Cohen's d < 0.1). Estimation errors increased with target angle magnitude in both movement directions. In young participants, estimation errors were non-symmetric, with estimations in flexion [ µ ( θ c o n y f ) = 1 . 8 ° , σ ( θ c o n y f ) = 7 . 0 ° ] being significantly more accurate than in extension [ µ ( θ c o n y e ) = 8 . 3 ° , σ ( θ c o n y e ) = 9 . 3 ° , p < 0.01]. This asymmetry was not present in the elderly group, where estimations in flexion [ µ ( θ c o n o f ) = 7 . 5 ° , σ ( θ c o n o f ) = 9 . 8 ° ] were similar to extension [ µ ( θ c o n o e ) = 7 . 7 ° , σ ( θ c o n o e ) = 9 . 3 ° ]. Hence, young and elderly participants performed equally in extension direction, whereas wrist position sense in flexion direction deteriorated with age (p < 0.01). Though unimpaired elderly adults did not benefit from stochastic stimulation, it cannot be deduced that individuals with more severe impairments of their sensory system do not profit from this treatment. While the errors in estimating wrist position are symmetric in flexion and extension in elderly adults, young adults are more accurate when estimating wrist flexion, an effect that has not been described before.

Stance and Swing Detection Based on the Angular Velocity of Lower Limb Segments During Walking.

Lower limb exoskeletons require the correct support magnitude and timing to achieve user assistance. This study evaluated whether the sign of the angular velocity of lower limb segments can be used to determine the timing of the stance and the swing phase during walking. We assumed that stance phase is characterized by a positive, swing phase by a negative angular velocity. Thus, the transitions can be used to also identify heel-strike and toe-off. Thirteen subjects without gait impairments walked on a treadmill at speeds between 0.5 and 2.1 m/s on level ground and inclinations between -10 and +10°. Kinematic and kinetic data was measured simultaneously from an optical motion capture system, force plates, and five inertial measurement units (IMUs). These recordings were used to compute the angular velocities of four lower limb segments: two biological (thigh, shank) and two virtual that were geometrical projections of the biological segments (virtual leg, virtual extended leg). We analyzed the reliability (two sign changes of the angular velocity per stride) and the accuracy (offset in timing between sign change and ground reaction force based timing) of the virtual and biological segments for detecting the gait phases stance and swing. The motion capture data revealed that virtual limb segments seem superior to the biological limb segments in the reliability of stance and swing detection. However, increased signal noise when using the IMUs required additional rule sets for reliable stance and swing detection. With IMUs, the biological shank segment had the least variability in accuracy. The IMU-based heel-strike events of the shank and both virtual segment were slightly early (3.3-4.8% of the gait cycle) compared to the ground reaction force-based timing. Toe-off event timing showed more variability (9.0% too early to 7.3% too late) between the segments and changed with walking speed. The results show that the detection of the heel-strike, and thus stance phase, based on IMU angular velocity is possible for different segments when additional rule sets are included. Further work is required to improve the timing accuracy for the toe-off detection (swing).

User-driven walking assistance: first experimental results using the MyoSuit.

Wearable robots for the legs have been developed for gait rehabilitation training and as assistive devices. Most devices have been rigid exoskeletons designed to substitute the function of users who are completely paralyzed. While effective for this target group, exoskeletons limit their users' contributions to movements. Soft wearable robots have been suggested as an alternative that allows, and requires, active contributions from users with residual mobility.In this work, we first tested if the MyoSuit, a lightweight, lower-limb soft wearable robot, affected the walking kinematics of unimpaired users. Secondly, we evaluated the assistance delivered to a patient with a gait impairment.In our first study, 10 unimpaired participants walked on a treadmill at speeds between 0.5 and 1.3 m/s. We found that wearing the MyoSuit in its transparency mode did not affect the participants' walking kinematics (RMS difference of joint angles < 1.6°). Step length and the ratio of stance-to-stride duration were not affected when wearing the MyoSuit.In our case study with one spinal cord injured participant, the MyoSuit supported the participant to increase his 10 MWT walking speed from 0.36 to 0.52 m/s, a substantial clinically meaningful improvement.Our results show that the MyoSuit allows user-driven, kinematically unaltered walking and provides effective assistance. Systems like the MyoSuit are a promising technology to bridge the gap between rigid exoskeletons and unassisted ambulation.

Making neurorehabilitation fun: Multiplayer training via damping forces balancing differences in skill levels.

Multiplayer environments are thought to increase the training intensity in robot-aided rehabilitation therapy after stroke. We developed a haptic-based environment to investigate the dynamics of two-player training performing time-constrained reaching movements using the ARMin rehabilitation robot. We implemented a challenge level adaptation algorithm that controlled a virtual damping coefficient to reach a desired success rate. We tested the algorithm's effectiveness in regulating the success rate during game play in a simulation with computer-controlled players, in a feasibility study with six unimpaired players, and in a single session with one stroke patient. The algorithm demonstrated its capacity to adjust the damping coefficient to reach three levels of success rate (low [50%], moderate [70%], and high [90%]) during singleplayer and multiplayer training. For the patient - tested in single-player mode at the moderate success rate only - the algorithm showed also promising behavior. Results of the feasibility study showed that to increase the player's willingness to play at a more challenging task condition, the effect of the challenge level adaptation - regardless of being played in single player or multiplayer mode - might be more important than the provision of multiplayer setting alone. Furthermore, the multiplayer setting tends to be a motivating and encouraging therapy component. Based on these results we will optimize and expand the multiplayer training platform and further investigate multiplayer settings in stroke therapy.

The Myosuit: Bi-articular Anti-gravity Exosuit That Reduces Hip Extensor Activity in Sitting Transfers.

Muscle weakness-which can result from neurological injuries, genetic disorders, or typical aging-can affect a person's mobility and quality of life. For many people with muscle weakness, assistive devices provide the means to regain mobility and independence. These devices range from well-established technology, such as wheelchairs, to newer technologies, such as exoskeletons and exosuits. For assistive devices to be used in everyday life, they must provide assistance across activities of daily living (ADLs) in an unobtrusive manner. This article introduces the Myosuit, a soft, wearable device designed to provide continuous assistance at the hip and knee joint when working with and against gravity in ADLs. This robotic device combines active and passive elements with a closed-loop force controller designed to behave like an external muscle (exomuscle) and deliver gravity compensation to the user. At 4.1 kg (4.6 kg with batteries), the Myosuit is one of the lightest untethered devices capable of delivering gravity support to the user's knee and hip joints. This article presents the design and control principles of the Myosuit. It describes the textile interface, tendon actuators, and a bi-articular, synergy-based approach for continuous assistance. The assistive controller, based on bi-articular force assistance, was tested with a single subject who performed sitting transfers, one of the most gravity-intensive ADLs. The results show that the control concept can successfully identify changes in the posture and assist hip and knee extension with up to 26% of the natural knee moment and up to 35% of the knee power. We conclude that the Myosuit's novel approach to assistance using a bi-articular architecture, in combination with the posture-based force controller, can effectively assist its users in gravity-intensive ADLs, such as sitting transfers.

A novel pneumatic stimulator for the investigation of noise-enhanced proprioception.

Executing coordinated movements requires that motor and sensory systems cooperate to achieve a motor goal. Impairment of either system may lead to unstable and/or inaccurate movements. In rehabilitation training, however, most approaches have focused on the motor aspects of the control loop. We are examining mechanisms that may enhance the sensory system to improve motor control. More precisely, the effects of stochastic subliminal vibratory tactile stimulation on wrist proprioception. We developed a device - based on a novel soft pneumatic actuator skin technology - to stimulate multiple sites simultaneously and independently. This device applies vibratory stimulation (amplitude < 0.50 mm, bandwidth 20-120 Hz) to the skin overlaying the tendons of a joint to target the receptors in charge of position and movement encoding. It achieves high spatial resolution (< 1 mm2), uses a soft and flexible interface, and has the potential to be used in combination with additional rehabilitation interventions. We conducted a feasibility study with 16 healthy subjects (11 younger - 6 females; 5 older - 2 females) in which a robotic manipulandum moved the subject's wrist to defined positions that had to be matched with a gauge. Comparing trials with and without stimulation we found that stochastic stimulation influenced joint position sense. The device we developed can be readily used in psycho-physical experiments, and subsequently benefit physiotherapy and rehabilitation treatments.

Evaluation of a mixed controller that amplifies spatial errors while reducing timing errors.

Previous results suggest that haptic guidance enhances learning of the timing components of motor tasks, whereas error amplification is better for learning the spatial components. In this paper we evaluate a novel mixed guidance controller that combines haptic guidance and error amplification to simultaneously promote learning of the timing and spatial components. The controller is realized using a saddle-like force field around the desired movement position. This force field has a stable manifold tangential to the trajectory that guides subjects in velocity related aspects. The force field has its unstable manifold perpendicular to the trajectory, which amplifies the normal (spatial) error. We conducted an experiment with twenty nine healthy subjects to test whether training with the mixed guidance controller resulted in better learning than training without guidance or with guidance-as-needed. Subjects trained two tasks: a continuous rhythmic task (circle) and a continuous single task (line). We found that the effectiveness of the training strategy depended on the task. Training with mixed guidance was especially beneficial for learning the timing components of the line, but limited learning of the circle. Perhaps the continuous change in the force directions during training of the circle was too difficult to interpret.

Robotic Rehabilitator of the Rodent Upper Extremity: A System and Method for Assessing and Training Forelimb Force Production after Neurological Injury.

Rodent models of spinal cord injury are critical for the development of treatments for upper limb motor impairment in humans, but there are few methods for measuring forelimb strength of rodents, an important outcome measure. We developed a novel robotic device--the Robotic Rehabilitator of the Rodent Upper Extremity (RUE)--that requires rats to voluntarily reach for and pull a bar to retrieve a food reward; the resistance of the bar can be programmed. We used RUE to train forelimb strength of 16 rats three times per week for 23 weeks before and 38 weeks after a mild (100 kdyne) unilateral contusion at the cervical level 5 (C5). We measured maximum force produced when RUE movement was unexpectedly blocked. We compared this blocked pulling force (BPF) to weekly measures of forelimb strength obtained with a previous, well-established method: the grip strength meter (GSM). Before injury, BPF was 2.6 times higher (BPF, 444.6 ± 19.1 g; GSM, 168.4 ± 3.1 g) and 4.9 times more variable (p < 0.001) than pulling force measured with the GSM; the two measurement methods were uncorrelated (R(2) = 0.03; p = 0.84). After injury, there was a significant decrease in BPF of 134.35 g ± 14.71 g (p < 0.001). Together, our findings document BPF as a repeatable measure of forelimb force production, sensitive to a mild spinal cord injury, which comes closer to measuring maximum force than the GSM and thus may provide a useful measure for quantifying the effects of treatment in rodent models of SCI.

Robot-assisted motor training: assistance decreases exploration during reinforcement learning.

Reinforcement learning (RL) is a form of motor learning that robotic therapy devices could potentially manipulate to promote neurorehabilitation. We developed a system that requires trainees to use RL to learn a predefined target movement. The system provides higher rewards for movements that are more similar to the target movement. We also developed a novel algorithm that rewards trainees of different abilities with comparable reward sizes. This algorithm measures a trainee's performance relative to their best performance, rather than relative to an absolute target performance, to determine reward. We hypothesized this algorithm would permit subjects who cannot normally achieve high reward levels to do so while still learning. In an experiment with 21 unimpaired human subjects, we found that all subjects quickly learned to make a first target movement with and without the reward equalization. However, artificially increasing reward decreased the subjects' tendency to engage in exploration and therefore slowed learning, particularly when we changed the target movement. An anti-slacking watchdog algorithm further slowed learning. These results suggest that robotic algorithms that assist trainees in achieving rewards or in preventing slacking might, over time, discourage the exploration needed for reinforcement learning.

Effort, performance, and motivation: insights from robot-assisted training of human golf putting and rat grip strength.

Robotic devices can modulate success rates and required effort levels during motor training, but it is unclear how this affects performance gains and motivation. Here we present results from training unimpaired humans in a virtual golf-putting task, and training spinal cord injured (SCI) rats in a grip strength task using robotically modulated success rates and effort levels. Robotic assistance in golf practice increased trainees feelings of competence, and, paradoxically, increased their sense effort, even though it had mixed effects on learning. Reducing effort during a grip strength training task led rats with SCI to practice the task more frequently. However, the more frequent practice of these rats did not cause them to exceed the strength gains achieved by rats that exercised less often at higher required effort levels. These results show that increasing success and decreasing effort with robots increases motivation, but has mixed effects on performance gains.