The seven articles in this special section focus on rehabilitation via bio-cooperative control. In this special section, the idea of bio-cooperative control has been applied to different kinds of rehabilitation devices (e.g., exoskeletal robots, end-effector robots, wheelchairs) used for the training of gait or arm movements or for the assessment of entire body activity of healthy subjects and patients with stroke or other neurological disorders.

Healthy individuals modulate muscle activation patterns according to their intended movement and external environment. Persons with neurological disorders (e.g., stroke and spinal cord injury), however, have problems in movement control due primarily to their inability to modulate their muscle activation pattern in an appropriate manner. A functionality test at the level of individual muscles that investigates the activity of a muscle of interest on various motor tasks may enable muscle-level force grading. To date there is no extant work that focuses on the application of exoskeleton robots to induce specific muscle activation in a systematic manner. This paper proposes a new method, named “individual muscle-force control” using a wearable robot (an exoskeleton robot, or a power-assisting device) to obtain a wider variety of muscle activity data than standard motor tasks, e.g., pushing a handle by hand. A computational algorithm systematically computes control commands to a wearable robot so that a desired muscle activation pattern for target muscle forces is induced. It also computes an adequate amount and direction of a force that a subject needs to exert against a handle by his/her hand. This individual muscle control method enables users (e.g., therapists) to efficiently conduct neuromuscular function tests on target muscles by arbitrarily inducing muscle activation patterns. This paper presents a basic concept, mathematical formulation, and solution of the individual muscle-force control and its implementation to a muscle control system with an exoskeleton-type robot for upper extremity. Simulation and experimental results in healthy individuals justify the use of an exoskeleton robot for future muscle function testing in terms of the variety of muscle activity data.

This paper presents the analysis of four psychophysiological responses in post-stroke upper extremity rehabilitation. The goal was to determine which psychophysiological responses would provide the most reliable information about subjects' psychological states during rehabilitation. Heart rate, skin conductance, respiration, and skin temperature were recorded in a stroke group and a control group during two difficulty levels of a pick-and-place task performed in a virtual environment using a haptic robot and during a cognitive task. Psychophysiological measurements were correlated with results of a self-report questionnaire. All four responses showed significant changes in response to the different tasks. Skin conductance differentiated between the two difficulty levels and was correlated with self-reported arousal in both stroke and control groups. Skin temperature differentiated between the two difficulty levels for the control group, but provided poor results for the stroke group. Heart rate and respiration increased during tasks, but their connection to psychological state was unclear. Results suggest that, of the four measured responses, skin conductance offers the most potential as a psychological state indicator, with other measures providing supplementary information. Psychophysiological measurements could thus be used in closed-loop biocooperative systems that would detect the user's psychological state and change the course of therapy accordingly.

Brain-computer interfaces (BCI) offer potential for individuals with a variety of motor and sensory disabilities to interact with their environment, communicate and control mobility aids. Two key factors which affect the performance of a BCI and its usability are the feedback given to the participant and the subject's motivation. This paper presents the results from a study investigating the effects of feedback and motivation on the performance of the Strathclyde Brain Computer Interface. The paper discusses how the performance of the system can be improved by behavior integration and human-in-the-loop design.

A physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals is proposed. First, in the time domain, the cardiac activity mean and the motion artifact noise of the ECG signal are modeled by a Hermite polynomial expansion and principal component analysis, respectively. A set of time domain accelerometer features is also extracted. A support vector machine (SVM) is employed for supervised classification using these time domain features. Second, motivated by their potential for handling convolutional noise, cepstral features extracted from ECG and accelerometer signals based on a frame level analysis are modeled using Gaussian mixture models (GMMs). Third, to reduce the dimension of the tri-axial accelerometer cepstral features which are concatenated and fused at the feature level, heteroscedastic linear discriminant analysis is performed. Finally, to improve the overall recognition performance, fusion of the multimodal (ECG and accelerometer) and multidomain (time domain SVM and cepstral domain GMM) subsystems at the score level is performed. The classification accuracy ranges from 79.3% to 97.3% for various testing scenarios and outperforms the state-of-the-art single accelerometer based PA recognition system by over 24% relative error reduction on our nine-category PA database.

We describe error-related potentials generated while a human user monitors the performance of an external agent and discuss their use for a new type of brain–computer interaction. In this approach, single trial detection of error-related electroencephalography (EEG) potentials is used to infer the optimal agent behavior by decreasing the probability of agent decisions that elicited such potentials. Contrasting with traditional approaches, the user acts as a critic of an external autonomous system instead of continuously generating control commands. This sets a cognitive monitoring loop where the human directly provides information about the overall system performance that, in turn, can be used for its improvement. We show that it is possible to recognize erroneous and correct agent decisions from EEG (average recognition rates of 75.8% and 63.2%, respectively), and that the elicited signals are stable over long periods of time (from 50 to $>600$ days). Moreover, these performances allow to infer the optimal behavior of a simple agent in a brain–computer interaction paradigm after a few trials.

While a large number of robotic exoskeletons have been designed by research teams for rehabilitation, it remains rather difficult to analyse their ability to finely interact with a human limb: no performance indicators or general methodology to characterize this capacity really exist. This is particularly regretful at a time when robotics are becoming a recognized rehabilitation method and when complex problems such as 3-D movement rehabilitation and joint rotation coordination are being addressed. The aim of this paper is to propose a general methodology to evaluate, through a reduced set of simple indicators, the ability of an exoskeleton to interact finely and in a controlled way with a human. The method involves measurement and recording of positions and forces during 3-D point to point tasks. It is applied to a 4 degrees-of-freedom limb exoskeleton by way of example.

To operate a wheelchair, people with severe physical disabilities may require assistance, which can be provided by robotization. However, medical experts report that an excess of assistance may lead to loss of residual skills, so that it is important to provide just the right amount of assistance. This work proposes a collaborative control system based on weighting the robot's and the user's commands by their respective efficiency to reactively obtain an emergent controller. Thus, the better the person operates, the more control he/she gains. Tests with volunteers have proven, though, that some users may require extra assistance when they become stressed. Hence, we propose a controller that can change the amount of support taking into account supplementary biometric data. In this work, we use an off-the-shelf wearable pulse oximeter. Experiments have demonstrated that volunteers could use our wheelchair in a more efficient way due to the proposed biometric modulated collaborative control.

This work introduces a hybrid brain–computer interface (BCI) composed of an imagery-based brain switch and a steady-state visual evoked potential (SSVEP)-based BCI. The brain switch (event related synchronization (ERS)-based BCI) was used to activate the four-step SSVEP-based orthosis (via gazing at a 8 Hz LED to open and gazing at a 13 Hz LED to close) only when needed for control, and to deactivate the LEDs during resting periods. Only two EEG channels were required, one over the motor cortex and one over the visual cortex. As a basis for comparison, the orthosis was also operated without using the brain switch. Six subjects participated in this study. This combination of two BCIs operated with different mental strategies is one example of a “hybrid” BCI and revealed a much lower rate of FPs per minute during resting periods or breaks compared to the SSVEP BCI alone ( ${rm FP}=1.46pm 1.18$ versus 5.40 $pm$ 0.90). Four out of the six subjects succeeded in operating the self-paced hybrid BCI with a good performance (positive prediction value ${rm PPVb}>$0.70).

The largest obstacles to signal transduction for electroencephalography (EEG) recording are the hair and the epidermal stratum corneum of the skin. In typical clinical situations, hair is parted or removed, and the stratum corneum is either abraded or punctured using invasive penetration devices. These steps increase preparation time, discomfort, and the risk of infection. Cross-linked sodium polyacrylate gel swelled with electrolyte was explored as a possible skin contact element for a prototype preparation-free EEG electrode. As a superabsorbent hydrogel, polyacrylate can swell with electrolyte solution to a degree far beyond typical contemporary electrode materials, delivering a strong hydrating effect to the skin surface. This hydrating power allows the material to increase the effective skin contact surface area through wetting, and noninvasively decrease or bypass the highly resistive barrier of the stratum corneum, allowing for reduced impedance and improved electrode performance. For the purposes of the tests performed in this study, the polyacrylate was prepared both as a solid elastic gel and as a flowable paste designed to penetrate dense scalp hair. The gel can hold 99.2% DI water or 91% electrolyte solution, and the water content remains high after 29 h of air exposure. The electrical impedance of the gel electrode on unprepared human forearm is significantly lower than a number of commercial ECG and EEG electrodes. This low impedance was maintained for at least 8 h (the longest time period measured). When a paste form of the electrode was applied directly onto scalp hair, the impedance was found to be lower than that measured with commercially available EEG paste applied in the same manner. Time-frequency transformation analysis of frontal lobe EEG recordings indicated comparable frequency response between the polyacrylate-based electrode on unprepared skin and the commercial EEG electrode on abraded skin. Evoked potential recordings demonstrated signa- - l-to-noise ratios of the experimental and commercial electrodes to be effectively equivalent. These results suggest that the polyacrylate-based electrode offers a powerful option for EEG recording without scalp preparation.

A rhesus monkey was trained to perform individuated and combined finger flexions of the thumb, index, and middle finger. Nine implantable myoelectric sensors (IMES) were then surgically implanted into the finger muscles of the monkey's forearm, without any adverse effects over two years postimplantation. Using an inductive link, EMG was wirelessly recorded from the IMES as the monkey performed a finger flexion task. The EMG from the different IMES implants showed very little cross correlation. An offline parallel linear discriminant analysis (LDA) based algorithm was used to decode finger activity based on features extracted from continuously presented frames of recorded EMG. The offline parallel LDA was run on intraday sessions as well as on sessions where the algorithm was trained on one day and tested on following days. The performance of the algorithm was evaluated continuously by comparing classification output by the algorithm to the current state of the finger switches. The algorithm detected and classified seven different finger movements, including individual and combined finger flexions, and a no-movement state $({hbox {chance performance}} = 12.5hbox{%})$ . When the algorithm was trained and tested on data collected the same day, the average performance was $43.8pm3.6hbox{%} n=10$. When the training-testing separation period was five months, the average performance of the algorithm was $46.5pm3.4hbox{%} { n}=8$. These results demonstrated that using EMG recorded and wirelessly transmitted by IMES offers a promising approach for providing intuitive, dexterous control of artificial limbs where human patients have sufficient, functional residual muscle following amputation.

In this paper, we analyze the correlations between four clinical measures (Fugl–Meyer upper extremity scale, Motor Activity Log, Action Research Arm Test, and Jebsen-Taylor Hand Function Test) and four robotic measures (smoothness of movement, trajectory error, average number of target hits per minute, and mean tangential speed), used to assess motor recovery. Data were gathered as part of a hybrid robotic and traditional upper extremity rehabilitation program for nine stroke patients. Smoothness of movement and trajectory error, temporally and spatially normalized measures of movement quality defined for point-to-point movements, were found to have significant moderate to strong correlations with all four of the clinical measures. The strong correlations suggest that smoothness of movement and trajectory error may be used to compare outcomes of different rehabilitation protocols and devices effectively, provide improved resolution for tracking patient progress compared to only pre- and post-treatment measurements, enable accurate adaptation of therapy based on patient progress, and deliver immediate and useful feedback to the patient and therapist.

This paper presents a novel computer-controlled bicycle ergometer, the TiltCycle, for use in human biomechanics studies of locomotion. The TiltCycle has a tilting (reclining) seat and backboard, a split pedal crankshaft to isolate the left and right loads to the feet of the pedaler, and two belt-driven, computer-controlled motors to provide assistance or resistance loads independently to each crank. Sensors measure the kinematics and force production of the legs to calculate work performed, and the system allows for goniometric and electromyography signals to be recorded. The technical description presented includes the mechanical design, low-level software and control algorithms, system identification and validation test results.

This paper presents a new, user-friendly, portable motion capture and gait analysis system for capturing and analyzing human gait, designed as a telemedicine tool to monitor remotely the progress of patients through treatment. The system requires minimal user input and simple single-camera filming (which can be acquired from a basic webcam) making it very accessible to nontechnical, nonclinical personnel. This system can allow gait studies to acquire a much larger data set and allow trained gait analysts to focus their skills on the interpretation phase of gait analysis. The design uses a novel motion capture method derived from spatiotemporal segmentation and model-based tracking. Testing is performed on four monocular, sagittal-view, sample gait videos. Results of modeling, tracking, and analysis stages are presented with standard gait graphs and parameters compared to manually acquired data.

We propose a new metric to assess robustness of the human postural control system to an impulsive perturbation (in this case, a mild backward impulse force at the pelvis). By applying concepts from robust control theory, we use the inverse of the maximum value of the system's sensitivity function (1/MaxSens) as a measure for robustness of the human postural control system, e.g., a highly sensitive system has low robustness to perturbation. The sensitivity function, which in this case is the frequency response function, is obtained directly using spectral analysis of experimental measurements, without need to develop a model of the postural control system. Common measures of robustness, gain and phase margins, however require a model to assess system robustness. To examine the efficacy of this approach, we tested thirty healthy subjects across three age groups: young (YA: 20–30 years), middle-aged (MA: 42–53 years), and older adults (OA: 71–79 years). The OA group was found to have reduced postural stability during quiet stance as detected by center of pressure measures of postural sway. The proposed robustness measure of 1/MaxSens was also found to be significantly smaller for OA than YA or MA $({ p}=0.001)$, implying reduced robustness among the older subjects in response to the perturbation. Gain and phase margins failed to detect any age-related differences. In summary, the proposed robustness characterization method is easy to implement, does not require a model for the postural control system, and was better able to detect differences in system robustness than model-based robustness metrics.