Keywords: Technology assessment, Neuroengineering, Neurological disease
Abstract: Quantifying upper limb impairment post-stroke is of essential importance to monitor motor recovery or to evaluate different therapeutic approaches. Instrumented assessments of upper limb function, such as the Virtual Peg Insertion Test (VPIT), often emulate a daily life manipulation activity that requires the subject to actively lift the arm against gravity, which can be challenging for severely impaired patients with arm weakness. With the aim of making the VPIT accessible to patients with severe arm weakness, we conducted a pilot study to analyze the feasability of combining this assessment with an arm weight support (AWS) device in 16 healthy subjects. Subjects performed the VPIT protocol without AWS device and with three different levels of weight support. Usability of combining the VPIT and the AWS device was high in healthy subjects: The VPIT could be successfully completed without collisions with the AWS device, the duration to set up the AWS device was on average 1.5 min, and subjects reported high levels of comfort while experiencing AWS. Metrics representing arm function were mostly not significantly influenced by the presence of the AWS device despite a decrease of 6.2% in movement smoothness, whereas grasping control was not significantly affected at all. The AWS level did not alter motor performance, even though subjects reported a decrease in perceived arm control with an increased AWS level. The high usability of combining the VPIT with an AWS device might enable the assessment of severely impaired patients in clinical practice. However, the influence of the AWS on outcome measures of the VPIT must be taken into account to make assessment results interpretable in the context of daily life reaching and manipulation situations without AWS.

Keywords: Mobility, Biomechanics, Technology assessment
Abstract: Gait disorders and subsequent falls are major causes of chronic disability. Most falls occur during walking when an individual fails to recover from a loss of balance. We have developed an innovative cable-driven robot that delivers unpredictable waist-pull perturbations while walking on a treadmill. In previous experiments, we showed that a training session with repeated waist-pulls induces acute motor adaptations in gait and balance recovery reactions. To date, it is unknown if the effects can be transferred to different types of external balance perturbations. This study aims to investigate if the exposure to repeated waist-pulls can improve the recovery reactions to slips. Fourteen healthy young adults were assigned to either an Experimental (EG) or a Control Group (CG). The EG was trained with multidirectional waist-pull perturbations of various intensities. The CG walked for a comparable amount of time on the treadmill with cables on, but without experiencing any waist-pull. Before and after the training, all participants were exposed to three slip-like perturbations by suddenly accelerating one belt of the treadmill in the forward direction. The Margin of Stability (MoS) was evaluated while reacting to the perturbations. At post-training, the EG showed increased stability in reaction to the waist-pull and slip-like perturbations. In contrast, the reaction to slip-like disturbances of the CG remained unchanged. A single session of gait training comprised of repeated waist-pulls showed immediate transfer to split-belt treadmill slips. The findings of this pilot study are encouraging for developing new fall prevention therapies.

Keywords: Technology assessment, Neurological disease
Abstract: Biofeedback of online system identification estimates of intrinsic and reflexive joint impedance can be used by able-bodied subjects to voluntarily modulate their reflexive impedance independent of the intrinsic contribution. Similar to EMG-based paradigms, this could potentially be used to reduce muscle hyperreflexia in people with spasticity by facilitating spinal neuroplasticity. However, it remains unanswered if spastic participants are able to use this specific feedback to modulate their reflexes. We show, while subjects were free to co-contract, that the system identification measures have a large linear association with independently measured and processed EMG measures. The impedance estimates were obtained using an existing algorithm with incremental improvements to increase general applicability and decrease bias on the identified parameters in both simulation an experimental data. The correlation with EMG-based measures demonstrates the validity of the use of joint impedance measures within a training paradigm to reduce hyperreflexia. This could potentially improve participant comfort, increase applicability across joints, target hyperreflexia at joint level and generate faster training effects.

Keywords: Technology assessment, Neurological disease
Abstract: Proprioception is a critical component of sensorimotor functions which directly affect recovery after neurological injuries. However, clinical tests of proprioception still lack sensitivity and reliability, while robotic devices can provide quantitative, accurate, and repeatable metrics. This work presents the analysis of the efficacy of a robotic assessment of wrist proprioception in terms of the capability to discern between movements along the different DoFs in a healthy population with a broad range of age. The effect of aging on the proprioceptive matching was analyzed to select an appropriate control group for the comparison with stroke patients, designed to confirm the hypothesis that a high percentage of stroke patients presents proprioceptive impairments in the acute and subacute states. Results show that the protocol is capable of detecting differences in performance along different movement directions, and that wrist proprioception does not deteriorate in the age ranges analyzed. Finally, stroke patients were less accurate in matching the position of their wrist, confirming the hypothesis that proprioceptive performance is often impaired in the acute and subacute phases of stroke.

Keywords: Force control, Neural networks, Prostheses control
Abstract: The force applied with a prosthetic device is fundamental for the correct handling of objects in daily tasks. However, it is also a factor that normally gets relegated to a secondary plane, as researchers mainly focus on decoding the user’s intent in terms of movements to be performed. Continuous estimates of the grasp force from the electromyographic (EMG) signals were proposed in the past. As motor actions are preplanned in humans, we hypothesized that it would be possible to decode the intended grasp force from the transient state of the EMG signal. We tested this hypothesis by using features extracted from surface HD-EMG recordings from forearm muscles, classified using artificial neural networks. Data from 6 able-bodied subjects were collected. They were trained and tested at segments of 120 ms with 20 ms overlap, starting 1 s before and ending 0.5 s after the detection of the onset with different subsets of channels. The results obtained showed that the transient phase contains information about the target grasp force, achieving predictions of 2.62 % MVC average absolute errors within 430 ms from the onset of the EMG.

Keywords: Neurological disease, Technology assessment, Neuroengineering
Abstract: There are currently no effective tools to assess the range of physical, cognitive, and social issues seen in the emerging HIV-stroke population. In turn, this poses a barrier to developing effective neurorehabilitation strategies to deal with the unique challenges of living with both HIV and stroke. Rehabilitation robotics provide a potential approach to address this problem. In this study, we develop a cognitive-motor task on the Haptic Theradrive, a single degree-of-freedom robot designed for upper limb rehabilitation. We collect preliminary data on healthy and HIV-stroke subjects from both upper limbs. We identify metrics that could potentially be useful in assessing motor and cognitive impairment across the HIV-stroke disability spectrum.

Keywords: Force control, Wearable and augmenting devices, Biomechanics
Abstract: Exoskeleton assisted gait training in children with cerebral palsy (CP) offers the potential to increase therapy dosage and intensity compared to current approaches. Here, we report the design and characterization of a pediatric knee exoskeleton for gait training outside of a clinical environment. A multi-layered closed loop control system and a microcontroller based data acquisition system were implemented to provide individualized control approaches and achieve device portability for home use. Step response tests show the averaged 90% rise time was 45 ms for 5 Nm, 35 ms for 10 Nm, 40 ms for 15 Nm. The gain-limited closed-loop torque bandwidth was 9Hz with a 9 Nm amplitude chirp in knee flexion and extension. The actuator has low output impedance (<0.5 Nm) at low frequencies expected during use. Future work will investigate the long term effects of providing children with CP knee extension assistance during daily walking on gait biomechanics with, and without, the device.

Keywords: Flexible instruments, Design, Dynamics and control
Abstract: This paper presents and experimentally evaluates a quasistatic mechanics-based model that describes the shape of concentric tube robotic (CTR) arms when they are eccentrically arranged along an also-flexible backbone. The model can estimate the shape of both the backbone and CTR arms, and can accommodate an arbitrary number of CTR arms arranged in an eccentric position with regards to the backbone’s neutral axis. Experimental evaluation with a prototype system on the benchtop highlights the promise of the end-to-end proposed modelling approach, as the error between model and experiment is around 11% of the manipulator’s overall length. The theory is the first step towards modeling multi-arm concentric tube robots, a class of continuum robots that is increasingly being considered for single-port surgical applications.

Keywords: Computer vision, Neural networks, Biomechanics
Abstract: A path along which the human knee joint moves can be estimated from real-time moving images or a sequence of static images. In case of many algorithms solving this problem, it is essential to locate the characteristic points (i.e., key-points) on each image and find the correspondence between them in the image sequence. In this paper we present an algorithm, which detects such key-points facilitating effective femur tracking in a sequence of X-ray images. We use a set of X-ray images manually labeled with the key-point positions, to train a Convolutional Neural Network (CNN) for the purposes of solving a regression task corresponding to finding key-point positions in previously unknown images. CNN hyper-parameters such as number of convolutions and layers, learning rate, regularization parameters, and activation functions were optimized using a tree of Parzen estimators guiding the process of training multiple models. Results for models with the best mean-square estimation error computed for a validation set and lowest structural complexity are presented. Key-point positions predicted by the CNN are on par with human predictions, even though the actual key-point position is ambiguous in some cases. The feasibility of detected key-points for femur tracking has been verified by several case studies.

Keywords: Micro- and nano-robotics, Biologically-inspired systems, Manipulation
Abstract: In this paper, we propose and evaluate a novel biological cell-based microrobot for applications in active targeting and treating of solid tumors. The microrobots are prepared by the phagocytosis of a mouse macrophage cell line with small size gold nanorods (sAuNRs) and anticancer drug (Doxorubicin) contained nanoliposomes (Dox-LPs). First, we synthesize sAuNRs with the average dimensions of 29 nm × 7 nm and high energy absorbance to 808 nm near-infrared (NIR) light. After that we prepare Dox-LPs having an average diameter of 145 nm and thermal sensitive property with over 65% of drug release after 60 min in a 43 oC and pH 5.5 condition. After that, we fabricate the microrobots by co-incubation of sAuNRs and Dox-LPs with the macrophages. Experiments with confocal microscopy show that the nanotherapeutics can be readily engulfed by the macrophages. Finally, we evaluate the targeting characteristics of the microrobots to mouse breast cancer (4T1) cells using a chemotactic test with a 3-D tumor spheroid resembling a solid tumor. Consequently, it is shown that the microrobots can successfully infiltrate into the tumor spheroid. Ultimately, the accompanying sAuNRs can be heated up to trigger the drug liberation from the thermal sensitive liposomes to kill microrobots and the tumor simultaneously.

Keywords: Micro- and nano-robotics
Abstract: This work presents an approach to realize soft microrobots with multiple flexible flagella using beaded-fibers driven via a periodic magnetic field. Paramagnetic iron oxide particles are embedded into the polymer matrix of electrospun beaded-fibers and form magnetism upon applying an external magnetic field. We demonstrate that the induced magnetization by an external magnetic field enables self-assembly of multiple adjacent beaded-fibers to form a microrobot with multiple flexible flagella. Frequency response of the assembled microrobot and the individual beaded-fibers is characterized experimentally, and shows that the propulsive force imparted to the fluid by the multiple flexible flagella increases the actuation frequency range of the microrobot and enhances its swimming speed. At relatively high actuation frequency (20 Hz), the average speed of the individual beaded-fiber is 0.11 body-lengthper-second, whereas the microrobot with multiple flagella swims at an average speed of 0.30 body-length-per-second. We also observe a slight difference in the swimming speed between the microrobot with multiple flexible flagella and its individual beaded-fibers at relatively low actuation frequencies.

Keywords: Soft robotics, Flexible instruments, Design
Abstract: Gallbladder diseases are the most common health problems in developed societies, affecting a large part of the adult population. Laparoscopic cholecystectomy is the elective surgical procedure for gallbladder pathologies. Despite the high number of procedures performed each year and the limited complication rate, the retraction of the liver remains the most important issue to be improved during a laparoscopic cholecystectomy. Being the gallbladder anatomically fixed under the liver, the surgeon has to move hepatic portions and has to stably retract this organ during the whole procedure. In order to overcome the limitations related to modern surgical tools for tissue retraction in minimally invasive procedures, we propose here an innovative system composed by a soft band that exploits the layer jamming phenomenon, for assuring a proper stiffness variation and guaranteeing a stable liver retraction. Design, material characterization (both external membrane and internal layer material) and ex-vivo assessment in simulated laparoscopic regime have been performed for validating this concept: promising results have been obtained, thus opening the way for further investigations.

Keywords: Design, Image-guided interventions
Abstract: The significance of the robots in the medical field have been increasing rapidly. Humans and robots working together increases the strengths and decreases the limitations of surgical operations. Human life makes safety the most important problem for medical robots, for which there are no universal standards. This paper presents detailed design methods for increasing medical robots' safety by considering issues of sterilization, robot's size, operating room placement of the robot, the robot mechanics, selection of the electromechanical components, drive mechanism, stiffness, sensor redundancy, software application, and hazard identification and analysis. The proposed safety design concepts were put into practice on a surgical robot prototype and the outcomes are discussed.

Keywords: Dynamics and control, Neuro robotics, Surgical navigation
Abstract: This paper proposes an approach for displaying the needle-tip interaction force exchanged between the needle tip and the tissues to the remote operator of a teleoperated needle insertion procedure. As known, the measures of the needle tip interaction force with tissues obtained through F/T sensor at the robot wrist do not provide a transparent perception of the needle-tissue interaction at the tip mainly because of the friction between the needle shaft and the traversed tissues. Current literature mainly proposes hardware solutions to the problem of measuring the forces at the needle tip. In this work we aim instead at cleaning the F/T sensor information for rendering only the estimated force exchanged at the needle tip. The approach is based on an online identification of the parameters of a needle-tissue interaction force model to isolate offset force values mainly due to friction. The approach, validated through simulations and experiments, is expected to increase the sensitivity of the rendered force to tissue transitions thus improving safety and accuracy in needle placement.

Keywords: Prostheses, Manipulation, Technology assessment
Abstract: Humans often use features of their environment for assistance in picking up and manipulating objects or in stabilizing their own bodies. This ‘exfordance’ use occurs when external contact or gravitational or inertial forces are utilized to aid in task completion or stabilization. This paper presents a categorization of exfordance use and applies the new framework to quantifying how experienced unilateral upper- limb amputees use of exfordances during everyday activities, both in their affected and unaffected limbs. Head-mounted cameras were used to record video footage of participants in their homes while they completed self-selected activities of daily living. A total of 35 minutes of dense manipulation footage has been analyzed for each of 5 trans-radial amputees with different prosthetic devices, resulting in over 4,700 instances of observed exfordance use. The results indicate that participants used exfordance-based vs. non exfordance-based manipulation strategies approximately the same amount with both their intact and prosthetic hands, after adjusting for overall hand use. Furthermore, the specific exfordance use strategies vary substantially between limbs, with participants using environmental surfaces such as tables to guide the motion of their unaffected hand more frequently than with their prosthetic hand, possibly due to increased control and passive conformation ability. Also, participants used gravity-based exfordances (e.g. hanging a towel over the hand) much more frequently with their prosthetic, likely due to its reduced grasping capabilities.

Keywords: Prostheses, Wearable and augmenting devices, Biologically-inspired systems
Abstract: Over the last decade, active lower-limb prostheses demonstrated their ability to restore a physiological gait for transfemoral amputees by supplying the required positive energy balance during daily life locomotion activities. However, the added-value of such devices is significantly impacted by their limited energetic autonomy and excessive weight preventing their full appropriation by the patients. There is thus a strong incentive to reduce both the overall power consumption and weight of active prostheses. To address these issues, we developed a novel parallel spring mechanism, tailored to the dynamical behavior of an ankle prosthesis. The first contribution is the development of a lightweight and adaptive locking system, comprising an energy efficient ratchet and pawl mechanism with electromagnetical actuation. As second contribution, the required compliance is directly materialized within the structure of the prosthesis with no additional parts, taking advantage of fused filament fabrication (FDM) technology with carbon fibers reinforcement. Our system provides an elastic torque during flat ground walking, corresponding to 41% of the peak torque produced by an healthy ankle 50Nm, at a negligible energetic cost 0.5J/stride. By design, the novel parallel spring mechanism is lightweight (140g), can engage at any plantarflexion position with a locking discretization of 0.3°, and is self-unlocking.

Keywords: Prostheses, Human-machine interaction, Biomechanics
Abstract: Most prosthetic feet behave like springs, and their stiffness affects many important facets of amputee gait. Despite the importance of prosthesis stiffness, the ability of amputees to sense stiffness changes—that is, distinguish between more or less stiff feet—is unknown. This perceptual resolution has implications for the methodology and overall significance of selecting the optimal foot stiffness during prescription. In this experiment, we used a custom, variable-stiffness ankle prosthesis to make small adjustments to stiffness in between steps, and below-knee amputees were asked to identify whether the ankle became more or less stiff. We determined that the average difference threshold of stiffness was 8%, meaning that subjects could correctly identify an 8% change in stiffness 75% of the time. This high sensitivity underscores the importance of optimizing prosthesis stiffness on an individual basis, and suggests a shift is needed in the characterization of commercial feet and the use of stiffness variation during the prescription process.

Keywords: Prostheses, Wearable and augmenting devices, Prostheses control
Abstract: Current developments in ankle prosthetics are focusing on integrated actuators to fully control torques and angles. This enables terrain adaptive strategies e.g. for stairs and ramps. EMG and motion sensor input are state of the art approaches to classify different terrain or terrain changes, but these approaches have limited capabilities and detection accuracy. We present a novel approach for the detection of obstacles using a wearable Frequency-Modulated Continuous Wave (FMCW) radar integrated into a lower limb prosthetic device. With the continuous rotational motion of the tibia during the swing and stance phase, the radar scans the profile of the terrain in sagittal plane in front of the prosthesis. Gait phases are detected using a neural network classifiers based on inertial sensor data. Performance of the system is demonstrated in a single stair detection scenario.

Keywords: Prostheses, Design and control, Wearable devices
Abstract: Challenges associated with current prosthetic technologies limit the quality of life of lower-limb amputees. Passive prostheses lead amputees to walk slower, use more energy, fall more often, and modify their gait patterns to compensate for the prosthesis’ lack of net-positive mechanical energy. Robotic prostheses can provide mechanical energy, but may also introduce challenges through controller design. Fortunately, talented researchers are studying how to best control robotic leg prostheses, but the time and resources required to develop prosthetic hardware has limited their potential impact. Even after research is completed, comparison of results is confounded by the use of different, researcher-specific hardware. To address these issues, we have developed the Open-source Leg (OSL): a scalable robotic knee/ankle prosthesis intended to foster investigations of control strategies. This paper introduces the design goals, transmission selection, hardware implementation, and initial control benchmarks for the OSL. The OSL provides a common hardware platform for comparison of control strategies, lowers the barrier to entry for prosthesis research, and enables testing within the lab, community, and at home.

Keywords: Prostheses, Wearable and augmenting devices, Prostheses control
Abstract: Though the human wrist greatly benefits manipulation by orienting the hand, recreating its functionality in anthropomorphic proportions proves to be difficult. As a result, upper limb amputees who desire wrist functionality must often use often passive single degree of freedom devices to attempt to recover some of this missing ability. In this work, the authors present the preliminary design and kinematic testing of a 3 DOF actuated wrist prosthesis. The device is capable of 90 degrees of circumduction (combined flexion and abduction), and continuous rotation in pronation. The circumducting portion of the wrist is composed of a parallel mechanism, while the pronation drive is a belt-driven serial mechanism. The circumducting DOFs are actuated via nonbackdrivable linear DC motors, and the gear ratio and friction of the pronation mechanism make it nonbackdrivable under reasonable loads as well. The architecture of this wrist was chosen such that when the device is integrated into a transradial socket, the motors could be placed on the socket periphery, allowing amputees with long residual limbs to use the wrist without significantly offsetting the terminal device. Benchtop evaluation of the wrist device is performed to evaluate the speed of the wrist in actuation from the neutral position to a variety of flexion and abduction positions, as well as to evaluate its pronation speed. We find the wrist performs rather near to the ideally expected speed values.

Keywords: Human-machine interfaces, Prostheses control, Biomechanics
Abstract: Upper limb loss substantially impacts on the quality of life of thousands of individuals worldwide. Current advanced treatments rely on myoelectric prostheses controlled by electromyograms (EMG). Despite advances in surgical procedures (i.e. targeted muscle reinnervation) as well as in electrode design and bio-electric signal sampling, current myocontrol schemes provide limited re-gain of functionality and lack of bio-mimesis. Current solutions create mappings between EMG and prosthesis joint angles, disregarding the underlying neuromusculoskeletal processes. The poor performance of these approaches determines high rejection rates (40-50%) of myoelectric bionic limbs. This paper presents a biomimetic paradigm for active prosthesis control. It encompasses a modelling formulation that simulates the amputee’s phantom limb musculoskeletal dynamics as controlled by high-density EMG-extracted neural activations to muscles. We demonstrate how this technique can be applied to a transhumeral amputee offline to decode musculoskeletal function in the phantom elbow and wrist offline. Moreover, we provide preliminary data showing how this technique can be operated online on intact-limbed individuals. The proposed paradigm represents an important step towards next-generation bionic limbs that can mimic human biological limb functionality and robustness.

Keywords: Prostheses, Prostheses control, Machine learning and reinforcement learning
Abstract: The Spike Sorting is an algorithm that allows extracting peculiar features from the neural signals and uniquely identifying the neurons that contributed to the generation of the recording. The literature shows that researches on this topic do not pay the due attention to the optimization process of the algorithm parameters. Here, an optimization process based on the multimodality approach is presented. It was aimed to select the best set of features to increase the accuracy of classification of neural signals. Simulated recordings were used to validate the approach. We demonstrated that triplets of optimized features were able to discriminate among 10 classes with an accuracy of ∼ 95%; on the other hand, a fixed triplet reached an accuracy of ∼ 90%. Moreover, accuracy decay with respect to the classes was slower and surprisingly more predictable.

Keywords: Human-machine interfaces, Machine learning and reinforcement learning, Design and control
Abstract: For many people suffering from motor disabilities, robotic devices are the only way to interact with their environment. Natural tasks often require different kinds of interactions, that would involve different robotic controllers the user should be able to switch between in a self-paced way. We developed a Brain-Computer Interface (BCI) allowing users to switch between four control modes in a self-paced way in real-time. Since the system is devised to be used in domestic environments in a user-friendly way, we selected non-invasive electroencephalographic (EEG) signals and convolutional neural networks (ConvNets), known for their ability to find the optimal features in classification tasks. We tested our system using the Cybathlon BCI computer game, which embodies all the challenges inherent to real-time control. Our preliminary results show that an efficient architecture (SmallNet), with only one convolutional layer can classify 4 mental activities chosen by the user. The BCI system is run and validated online. It is kept up-to-date through the use of newly collected signals along playing, reaching an online accuracy of 47.6% where most approaches only report results obtained offline.

We found that models trained with data collected online better predicted the behaviour of the system in real-time. This suggests that similar (ConvNets based) offline classifying methods found in the literature might experience a drop in performance when applied online. Compared to our previous decoder of physiological signals relying on blinks, we increased by a factor 2 the amount of states the accuracy among which the user can transit, bringing the opportunity for finer control of specific subtasks composing natural grasping in a self-paced way.

Keywords: Neural networks, Biologically-inspired systems, Neuro robotics
Abstract: The cerebellum has a central role in fine motor control and in various neural processes, as in associative paradigms. In this work, a bioinspired adaptive model, developed by means of a spiking neural network made of thousands of artificial neurons, has been leveraged to control a humanoid NAO robot in real-time.

The learning properties of the system have been challenged in a classic cerebellum-driven paradigm, the Pavlovian timing association between two provided stimuli, here implemented as a laser-avoidance task. The neurophysiological principles used to develop the model, succeeded in driving an adaptive motor control protocol with acquisition and extinction phases.

The spiking neural network model showed learning behaviors similar to the ones experimentally measured with human subjects in the same conditioning task. The model processed in real-time external inputs, encoded as spikes, and the generated spiking activity of its output neurons was decoded, in order to trigger the proper response with a correct timing. Three long-term plasticity rules have been embedded for different connections and with different time-scales. The plasticities shaped the firing activity of the output layer neurons of the network.

In the Pavlovian protocol, the neurorobot successfully learned the correct timing association, generating appropriate responses. Therefore, the spiking cerebellar model was able to reproduce in the robotic platform how biological systems acquire and extinguish associative responses, dealing with noise and uncertainties of a real-world environment.

Keywords: Human-machine interaction, Wearable and augmenting devices, Human-machine interfaces
Abstract: Motion intent recognition is an important part of autonomous rehabilitative and assistive devices. The focus of this paper is on upper limb motion intent detection for use in wearable rehabilitative devices. Our aim is to capture the tactile cues that arise during weak muscle contractions. We introduce the tactile arm brace (TAB) and analyse the various patterns recorded. Depending on the arm muscles that contract to perform a particular hand motion, different parts of the TAB will experience variations in the interaction forces. A study involving 12 healthy subjects was conducted using the TAB and a bespoke gripping device, designed and built to measure gripping forces. Tactile signatures on the brace change when the fingers extend or grip; this validates our hypothesis based on the forearm muscle physiology. Gripping models were generated to relate low-strength gripping, between 0kg and 2kg, and TAB force readings. The best recorded sensitivity (ratio between the gripper force and the FSR), found in the proximity of the posterior radial forearm, was 0.052. Using all TAB sensors, the K-fold cross validation method produced an RMSE of 5.48N.

Keywords: Prostheses control, Human-machine interfaces, Design and control
Abstract: Dexterous hand prostheses controlled via surface electromyography represent the most advanced non invasive functional restorative solution for hand amputees. However, control difficulties, comfort problems and high costs are still the main limitations of such commercial devices. The high cost can represent a barrier that is difficult to overcome, especially for pediatric populations and in developing countries. Low-cost technology was successfully used in the hand prosthetics field in recent years. In previous work, a low-cost gesture recognition armband called Myo showed promising results for hand gesture classification tasks in intact subjects. Most of these applications were based on machine learning techniques applied to the Myo raw data. However, the classifier provided with the Myo is able to identify five hand gestures, providing capabilities as a myoelectric control system. No studies have quantitatively investigated its performance in subjects with hand amputation, yet. The aim of this study is to quantitatively evaluate the performance of the Myo hand gesture classifier in hand amputees. Three subjects with hand amputation were asked to attempt performing the five pre-set hand gestures. Each gesture was repeated three times with the arm in three different postures. The subjects did not perform any training and did not receive any feedback. Overall classification accuracy for the four hand gestures based on electromyographic data ranged between 50% and 97%. A clear relation between the length of the residual limb and the classification accuracy was observed. The results show that the Myo built-in classifier can provide good performance when tested on hand amputees, supporting its applicability as a low-cost myoelectric control system.

Keywords: Human-machine interfaces, Wearable devices, Neural networks
Abstract: The human machine interface (HMI) refers to a paradigm in which the users interact with external devices through an interface that mediates the information exchanges between them and the device. In this work we focused on an HMI that exploits signals derived from the body to control the machine: the body machine interface (BMI). It is reasonable to assume that signals derived from body movements, electromyography activity, as well as brain activity, have a non-linear nature. This implies that linear algorithms cannot exploit all the information contained in these signals. In this work we proposed a new BMI that maps electromyographic signals into the control a computer cursor by using a new non-linear dimensionality reduction algorithm based on autoassociative neural network. We tested the system on a group of eight healthy subjects that, controlling this cursor, performed a reaching task. We compared the result with the performance of a age and gender matched group of healthy subjects that solved the same task using a BMI based on a linear mapping. The analysis of the performance indices showed a substantial difference between the two groups. In particular, the performance of the people using the non-linear mapping were better in terms of time, accuracy and smoothness of the cursor’s movement. This study opened the way to the exploitation of non-linear dimensionality reduction algorithms to pursue a new and effective clinical approach for body-machine interfaces.

Keywords: Technology assessment, Neurological disease
Abstract: One of the major issues for the success of assistive robotics concerns the question whether patients not only accept the technology and profit from it, but also whether they can effectively use it. This is especially relevant when patients are highly impaired and present several functional limitations. Therefore, it is important to enable patients to control robots with alternative methods during their activities of daily living. This work deals with the development of a voice control system based on the ROS middleware framework. The voice control was customized on the JACO2 (Kinova Technology, Montreal, QC, Canada), a 6 degree-of-freedom assistive robotic manipulator. Two subjects with different impairments due to a neurodegenerative disease tested the robot for usability controlling it through the JACO2 joystick and the developed voice control system. Both subjects used the voice control system successfully and scored highly its usability. The most impaired subject preferred the voice control while, by contrast, the less impaired one preferred to use the joystick. Preliminary results showed good usability of the system, which could be an important aid for highly impaired people.

Keywords: Machine learning and reinforcement learning, Wearable devices, Exoskeletons
Abstract: Stroke affects the mobility, hence the quality of life of people victim of this cerebrovascular disease. Exoskeletons are developed to bring support to the user’s joints in order to improve their gait and to help regaining independence in daily life. Among them, Xosoft is a soft modular exoskeleton currently under research in the framework of the European project of the same name. On top of its assistive properties, it will provide therapeutic feedback via the analysis of kinematic data stemming from inertial sensors mounted on it. However, the activities performed by the user must be known beforehand in order to have sufficient behavioral context to interpret this data. In this study, four activity recognition chains, based on machine learning algorithm, were developed to automatically identify the nature of these activities. To be consistent with the application they are being used for, focus was made on reducing energy consumption by configuration minimization and bringing robustness to these algorithms. Movement sensor data was collected on eleven stroke survivors while performing daily-life activities. Influence of sensor reduction and position on the performances of the algorithms was evaluated and their resistance to sensor failures was assessed. Results show that in all four chains, and for each patient, reduction of sensors is possible until a certain limit beyond which the position has to be carefully chosen to maintain good performances. In particular, the study shows the benefits of avoiding lower legs and foot locations as well as the sensors positioned on the affected side of the user. It also shows that robustness can be brought to the chain when the data stemming from the different sensors are fused at the very end of the classification process.

Keywords: Design and control
Abstract: Functional Electrical Stimulation (FES) systems are successful in restoring motor function and supporting paralyzed users. But, current commercially available FES products are open loop, meaning that the system is unable to adapt to changing conditions with the user and their muscles which results in muscle fatigue and poor stimulation protocols. This is because it is difficult to close the loop between stimulation and monitoring of muscle contraction using adaptive stimulation, because FES causes electrical artefacts which make it challenging to monitor muscle contractions with traditional methods such as electromyography (EMG). We overcome this limitation by combining FES with novel mechanomyographic sensors (MMG) to be able to monitor muscle activity during stimulation in real time. To provide a meaningful task we built a FES cycling rig with software interfaces that enable us to perform adaptive recording and stimulation and combine this with sensors on forces applied to the pedals using force sensitive resistors (FSRs); crank angle position using a magnetic incremental encoder and inputs from the user using switches and a potentiometer. We illustrated this with a closed-loop stimulation algorithm that used the inputs from the sensors to control the output of a programmable RehaStim 1 FES stimulator (Hasomed) in real-time. A recumbent bicycle rig was used as a testing platform for FES cycling. The algorithm was designed to respond to a change in requested speed (RPM) from the user and change the stimulation parameters until this speed was achieved and then maintain it.

Keywords: Machine learning and reinforcement learning, Human-machine interfaces, Prostheses control
Abstract: The interest on wearable prosthetic devices has boost the research for a robust framework to help injured subjects to regain their lost functionality. A great number of solutions exploit physiological human signals, such as Electromyography (EMG), to naturally control the prosthesis, reproducing what happens in the human limbs. In this paper, we propose for the first time a way to integrate EMG signals with Inertial Measurement Unit (IMU) information, as a way to improve subject-independent models for controlling robotic hands. EMG data are very sensitive to both physical and physiological variations, and this is particularly true between different subjects. The introduction of IMUs aims at enriching the subject-independent model, making it more robust with information not strictly dependent from the physiological characteristics of the subject. We compare three different models: the first based on EMG solely, the second merging data from EMG and the 2 best IMUs available, and the third using EMG and IMUs information corresponding to the same 3 electrodes. The three techniques are tested on two different movements executed by 35 healthy subjects, by using a leave-one-out approach. The framework is able to estimate online the bending angles of the joints involved in the motion, obtaining an accuracy up to 0.8634. The resulting joint angles are used to actuate a robotic hand in a simulated environment.

Keywords: Wearable devices, Machine learning and reinforcement learning, Exoskeletons
Abstract: Recording 3D ground reaction forces through instrumented crutches can assist patients undergoing ambulatory rehabilitation as well as help roboticists develop new assistive controllers for their exoskeletons. Current methods to measure the amount of weight a patient exerts on their limbs are either inaccurate, or not feasible outside of ideal laboratory conditions. This paper introduces Smart Crutches, an instrumented crutch system capable of measuring the weight that a patient places on his/her lower extremities and providing vibratory feedback in response to the measured weight. The device was calibrated using a motion capture system and force plates. Linear regression and support vector regression (SVR) were used for calibration, and 10-fold cross-validation was applied to estimate the system’s accuracy. Results indicate that machine learning regression methods may lead to improved accuracy, but the choice of the kernel function is critical. Gaussian kernel yielded root-mean-square errors (RSME) of 2.5N or less relative to force plates, while other kernel functions produced more inconsistent and less accurate results. Instrumented crutches may be a valid alternative to force plates for estimating ground reaction forces in crutch gait.

Keywords: Prostheses control, Machine learning and reinforcement learning, Human-machine interfaces
Abstract: Muscle synergies in humans are context-dependent---they are based on the integration of vision, sensorimotor information and proprioception. In particular, visual information plays a significant role in the execution of goal-directed grasping movements. Such contextual synergies are largely absent from modern prosthetic robots. In this work, we therefore introduce a new algorithmic contribution to support the context-aware, synergistic control of multiple degrees-of-freedom of an upper-limb prosthesis. In our previous work, we showcased an actor-critic reinforcement learning method that allowed someone with an amputation to use their non-amputated arm to teach their prosthetic arm how to move through a range of coordinated motions and grasp patterns. We here extend this approach to include visual information that could potentially help achieve context-dependent movement. To study the integration of visual context into coordinated grasping, we recorded computer vision information, myoelectic signals, inertial measurements, and positional information during a subject's training a robotic arm. Our approach was evaluated via prediction learning, wherein our algorithm was tasked with accurately distinguishing between three different muscle synergies involving similar myoelectric signals based on visual context from a robot-mounted camera. These preliminary results suggest that even simple visual data can help a learning system disentangle synergies that would be indistinguishable based solely on motor and myoelectric signals recorded from the human user and their robotic arm. We therefore suggest that integrating learned, vision-contingent predictions about synergies into a prosthetic control system could potentially allow prostheses to better adapt to daily-life situations.

Keywords: Prostheses control, Human-machine interfaces, Design and control
Abstract: Assistive robotics for human grasp support are either in the form of tele-operated robotic grippers or act through orthotic control of a paralyzed user's hand. Such devices require correct orientation for successful and efficient grasping. In many human-robot assistive settings, the end-user is required to explicitly control the many degrees of freedom of the robotic assistance making effective or efficient control problematic. Here we are demonstrating the off-loading of low-level control of assistive robotics and orthotics, through automatic end-effector orientation control for grasping. This paper describes a compact algorithm implementing computer vision techniques to obtain the orientation of the target object to be grasped, by segmenting the images acquired with a camera positioned on top of the end-effector of the robotic device. The rotation needed for optimized grasping is directly computed from the object's orientation. The algorithm has been evaluated in 6 different image backgrounds and approaching conditions, with 26 objects of different properties. 94.8% of the items were detected in all backgrounds, 56.3% achieving complete detection and 2.6%, 16.0% and 19.4% being affected by shadows, illumination or grayscale parts respectively. The grasping of the object is achieved for 91.1% of the cases and has been evaluated with a robot simulator confirming the promising performance of the algorithm.

Keywords: Prostheses, Haptics, Wearable devices
Abstract: It has been reported in the literature that sensory information is a valuable and desired form of feedback for prosthetic users. Communication of how the arm moves can reduce cognitive load, reduce the need for visual attention and help the user predict the initial grasping force. In this paper, a new method of communicating movement sensations is presented through the application of tactile apparent movement. By overlapping vibration created by arrays of linear resonant actuators, a stroking movement can be felt on the user’s arm. The results show potential for a low cost and light weight system that can communicate stimulations for up to three degrees of actuation in a prosthetic.

Keywords: Haptics, Wearable devices, Prostheses control
Abstract: We designed a wearable haptic feedback system to assist stair descent for users of lower limb prostheses. Our haptic feedback system consists of a custom insole with four force sensors, a thigh band with four vibrotactile actuators, and an on-board embedded processor. The loss of sensation and range of motion in the foot-ankle complex presents a challenge to lower limb prosthesis users during stair descent. By providing force information through haptic feedback, we restore access to sensory information of the stair edge.

We tested our system on 2 subject experiments. In the first experiment, 15 subjects wearing the thigh band were tasked to report the perceived position of stimulation on the sensor insole. All subjects were able to detect the stimulation position in terms of the actuator that was vibrating with a minimum of 82% accuracy for all four positions. In the second experiment, 13 subjects wore the haptic feedback system and the sensor insole and stepped on a visually obstructed staircase step of variable depth. Overall, the provision of haptic feedback reduced error in reporting the step edge position. Our haptic feedback system demonstrates the capability to provide sensory information that is otherwise absent, enabling a more natural stair descent. Future work includes performing the experiments on subjects with lower limb amputation and a complete stair descent performance assessment wearing the haptic feedback system.

Keywords: Haptics, Neurological disease, Technology assessment
Abstract: Touch provides important information on the physical properties of external objects, and contributes to the sense of our hand position and displacement in perceptual tasks. Recent studies showed that the texture of the touched surface produced a bias on the perceived tactile motion, ultimately affecting the direction of hand motion in reaching tasks. Specifically, moving on a plate with parallel ridges, the hand motion deviates towards a direction opposite with respect to the one predicted by tactile flow mathematical model, i.e. perpendicular to the ridges. Here, we used this phenomenon to quantitatively assess an impairment in tactile channel. We asked healthy participants slide the hand on a plate with parallel ridges, either with bare fingertip or by wearing a glove. The glove condition simulated a dysfunction in tactile channel, as may occur in pathological conditions, for e.g. due to a neurological disease. Our hypothesis is that, wearing a glove, the systematic error induced by the texture orientation will be smaller because the information provided by the tactile channelisnoisier. Results are in agreement with our hypothesis, and could open interesting perspectives towards a quantitative technology-based tool for the assessment of tactile impairment in pathological conditions.

Keywords: Prostheses, Haptics, Wearable devices
Abstract: An effective method of communicating sensory feedback for prosthetics is presented using a combination of mechanical pressure and skin stretch, resulting in a mixture of normal and shear force being applied to the human arm. Stimulations were induced on the subject’s forearm by three mechanical cranks, each attached to their own servo motor. Three different crank orientations were tested, each producing a different skin stretch direction, with the results showing that shear force/tangential skin stretch applied longitudinally to the forearm was perceived more easily as it produced the best recognition rate. With minimal training, eighteen able-bodied test subjects were able to recognise six different grips with an accuracy of up to 88%, and achieved an accuracy of 80% when recognising the six grips at two different pressure levels. This sensory feedback mechanism shows potential for a simple, easy to learn stimulation device that could help improve users control and embodiment of their prosthetic device that requires three separate feedback channels.

Keywords: Novel actuators, Human-machine interfaces, Design and control
Abstract: An electronic travel aid is an assistive device for people with visual impairments which comprised of a sensor and feedback display to extend sensing range beyond the range of a conventional white cane and present augmented feedback. While various electronic travel aids have been developed, the conventional white cane still remains as the most frequently used mobility tool. In this paper we introduce a new, small and lightweight haptic pin display driven by a single actuator based on a sliding plate or camshaft mechanism. The filtered information of obstacles is obtained via a LIDAR sensor and presented through the proposed pin display. Five different rotational frequencies of the motor shaft correspond to five distance levels. Together with an ultrasonic sensor for detecting head level obstacles and a piezo speaker for alarming obstacles at the head level, a prototype of an electronic travel aid was implemented. Further, ten subjects participated in the distance level identification study by utilizing the developed electronic travel aid. Result shows about 85% accuracy rate in the identification of five distance levels.

Keywords: Prostheses control
Abstract: Powered lower limb prostheses have the capabilities to assist individuals with a lower limb amputation during ambulation. While these devices can generate power at the knee and/or ankle to assist with incline walking and stair climbing, it is difficult to control the transition between these ambulation modes in a seamless and natural way. Pattern recognition has been suggested as an alternative to using a key fob to switch between modes and recent results have shown reliable performance (less than 5% error rate) across five ambulation modes. In this study we investigated performance of a similar system across multiple sessions of use, a necessary step prior to clinical use. Two individuals with a transfemoral amputation used a powered knee-ankle for five ambulation activities including level-ground walking, ramp ascent, ramp descent, stair ascent, and stair descent over four sessions spaced out over at least two months. An intent recognition system was trained using embedded prosthesis mechanical sensors with varying amounts of data collected across the sessions to determine the effect of multi-session use and increased variation in the activities trained. Overall system error rate decreased from 1.45% [0.3%] when the system was trained with Session 1 data only and tested with Session 4 data to 0.60% [0.02%] when the system was trained with Sessions 1-3 data and tested with Session 4 data. These results demonstrate that a reliable intent recognition system can be created with multiple sessions of use, bringing lower limb intent recognition systems for powered prostheses one step closer to clinical viability.

Keywords: Force control, Prostheses control, Haptics
Abstract: Amputees living in the developing world can benefit greatly from a dexterous low-cost robotic prosthetic hand that can be controlled via electromyography (EMG). This research addresses part of the challenge of designing and constructing such a low-cost device. In particular, the development of novel and functionally suitable fingertip sensors is presented in this paper. The sensors allowed for the user with a trans-humeral amputation to intuitively control grip strength of the robotic prosthetic hand with the help of an EMG electrode placed on the bicep muscle, as well as, a haptic sensory feedback system. The fingertip sensors illustrated a stable linear relationship with force, an even sensitivity to force over the pulp of finger and the medial and lateral sides of the finger above the distal inter-phalangeal joint across the fingertip. Additionally, it had a low cost of construction (1.00) and the ability to fit on curved surfaces. Two test subjects evaluated the performance of the sensors in combination with the haptic sensory feedback system. The use of the novel sensors allowed for the test subjects to discriminate the forces experienced by each finger when gripping objects of different shapes, with an accuracy of 80% and 73% accuracy respectively. Hence, the fingertip sensors along with haptic feedback can provide a possible solution for amputees to regain the sense a touch and at a low cost. This is a step towards a cost effective (150), yet functional robotic prosthetic hand for amputees.

Keywords: Wearable devices, Machine learning and reinforcement learning, Mobility
Abstract: Goal: Textile-based stretch sensors are a novel and innovative alternative to traditional wearable sensors to monitor human mobility and health status. However, due to their non-linear properties it can be challenging to obtain accurate information. The goal of this study was to investigate if machine learning can be applied to obtain more accurate measurements.

Methods: In a tensile test using a linear stage setup, data were collected from two commercial available stretch sensors (Adafruit and Image SI) and one self-fabricated sensor (Simon Fraser University, Canada). For each sensor, one hour of consecutive stretches in both a trapezoidal and sinusoidal input pattern were collected. We identified a set of features, trained three commonly used machine learning algorithms, and compared their performance in estimating the amount of stretch. To demonstrate the feasibility of our approach in real life, we tested our setup in two human applications. First, we attached a stretch sensor to the human chest to estimate the expansion of the rip cage during breathing. Second, we evaluated the performance in estimating the ankle position with a sensor attached to the foot.

Results: In the tensile test, Support Vector Regression performed best with an average accuracy (R2) of 0.98 (0.01) and mean absolute error of 0.18 (0.06) mm across all input patterns and sensors. An accuracy (R2) of 0.91 (0.04) with a mean absolute error of 3.08 (0.38) mm has been achieved in estimating the expansion of the chest. Similarly, an accuracy (R2) of 0.90 (0.04) with a mean absolute error of 2.90 (0.61) degree has been achieved in estimating the ankle position.

Conclusion: We demonstrate that machine learning can be used to obtain accurate stretch information from textile-based stretch sensors.

Keywords: Neuro robotics, Dynamic vision sensors, Neuroengineering
Abstract: Prediction is believed to play an important role in the human brain. However, it is still unclear how predictions are used in the process of learning new movements. In this paper, we present a method to learn movements from visual prediction. The method consists of two phases: learning a visual prediction model for a given movement, then minimizing the visual prediction error. The visual prediction model is learned from a single demonstration of the movement where only visual input is sensed. Unlike previous work, we represent visual information with event streams as provided by a Dynamic Vision Sensor. This allows us to only process changes in the environment instead of complete snapshots using spiking neural networks. By minimizing the prediction error, movements visually similar to the demonstration are learned. We evaluate our method by learning simple movements from human demonstrations on different simulated robots. We show that the definition of the visual prediction error greatly impacts movements learned by our method.

Keywords: Haptics, Biomechanics, Force control
Abstract: Motor learning or motor adaptation is the capability to acquire new motor skills or the adaptation of existing motor skills to new environmental conditions. In this paper, a new protocol based on a low-cost haptic device for evaluating the motor adaptation during perturbed 3D reaching tasks was presented. The protocol consisted of three 3D reaching tasks performed using Novint Falcon: a familiarization task in which no force field was applied, an adaptation task in which a perturbing force field occurred, and a wash out task with no force field. Ten healthy subjects were enrolled in the study. Subjects were asked to reach four targets equally distributed along a circumference. During the adaptation task, a constant force perpendicular to the direction of movement was applied and it was randomly removed 40 times out of 160. Trajectories of the end-effector were recorded to calculate the following kinematic indices: duration of movement, length ratio, lateral deviation, speed metric and normalized jerk. The learning index was calculated to study the motor learning during the adaptation task. Two-way repeated measure ANOVA tests were performed for all the indices considering movement directions and tasks as independent variables. Moreover, a one-way repeated measure ANOVA was performed on the learning index to find differences among the 4 target sets. The movement accuracy is influenced from both the perturbed force field and the movement direction. The smoothness of the reaching movement is influenced by the presence of the force field and decreases when it is applied. Learning index showed the capability of the subjects to rapidly adapt to a perturbed force field, generating a compensation strategy in a 3D movement.

Keywords: Biomechanics, Haptics, Human-machine interaction
Abstract: During movement, our central nervous system (CNS) takes into account the dynamics of our environment to optimally adapt our joint dynamics. In this study we explored the adaptation of shoulder joint dynamics when a participant interacted with a time-varying virtual environment created by a haptic manipulator. Participants performed a position task, i.e., minimizing position deviations, in face of continuous mechanical force perturbations. During a trial the environmental damping, mimicked by the manipulator, was either increased (0 to 200 Ns/m) or decreased (200 to 0 Ns/m) in 1 s or 8 s. A system identification technique, kernel-based regression, was used to reveal time-varying shoulder joint dynamics using the frequency response function (FRF). The FRFs revealed that the rate at which shoulder joint dynamics is adapted depends on the rate and direction of change in environmental damping. Adaptation is slow, but starts immediately, after the environmental damping increases, whereas adaptation is fast but delayed when environmental damping decreases. The results obtained in our participants comply with the framework of optimal feedback control, i.e., adaptation of joint dynamics only takes place when motor performance is at risk or when this is energetically advantageous.

Keywords: Haptics, Neurological disease, Wearable devices
Abstract: Virtual reality-based mirror therapy has been suggested as an option for the rehabilitation of stroke patients. One advantage of using virtual environments is the option to introduce scaling between motions in the real and the virtual world. However, such mismatches only remain unnoticed by a user up to a certain threshold. In this work, we address the effect of providing haptic stimuli on such perceptional thresholds. Specifically, we focused on rendering tactile taps corresponding to object contacts during the abduction of the upper extremities. In a user study, we confirmed our conjecture that providing haptic feedback increases the scaling threshold, at which the aforementioned differences become noticeable. These findings have direct relevance for the development of virtual reality-based motion-oriented stroke therapies.

Keywords: Biomechanics, Human-machine interfaces, Prostheses control
Abstract: This study investigated artificial neural networks (ANN) that quantified the loaded ankle impedance of healthy subjects as they contracted their lower extremity muscles. The multivariable standing ankle impedance of 12 male subjects was determined using an instrumented vibrating platform. Surface electromyography (EMG) was used to measure the muscle activity while the subject’s muscles were relaxed, and co-contracted to 10%, 20%, 30%, and 40% of the subject’s maximum voluntary contraction (MVC). The function fitting capabilities of ANN were used to relate the input information (measured EMG and subject biometrics) to the desired output ankle impedance parameters (stiffness, damping, and inertia). The results showed that the relationship between muscle activity and standing ankle impedance can be modeled with high accuracy, and showed feasibility towards a generalized model that works for subjects that did not participate in the experiment. Using the predicted impedance, future work could investigate this relationship during walking and be used to advance active ankle-foot prostheses control based on the user’s intentions.

Keywords: Neuro robotics, Locomotion and manipulation in robots and biological systems, Neuroengineering
Abstract: Robot control is an active field of research, in particular humanoid robots present challenging problems. Yet, very few of the proposed methods exhibit properties of biological systems. Neurorobotics presents an interesting approach by using neural models and mechanisms of brain function to control robots. In this paper, we present a novel way to activate robot motions with different modalities, inspired by biology and implemented with spiking neurons. We focus on two specific characteristics of biological systems for motion control: the hierarchical representation, which is distributed in the body and the nervous system, and the different activation modalities. We modeled three different activation modalities: voluntary, rhythmic and reflexes. In our architecture, motions are represented with motor primitives that can be combined and parameterized. A mechanism to learn new motions based on previous knowledge is incorporated using an error function. Our approach is evaluated in different scenarios combining the different modalities in various ways. We demonstrate the method by controlling a robotic arm in simulation, and by learning new trajectories as primitives.

Keywords: Exoskeletons, Wearable and augmenting devices, Technology assessment
Abstract: Robotic rehabilitation devices have gained significant popularity in the past decade. Over-ground leg exoskeletons commonly use traditional rigid link architectures or support the weight of a user by strapping the user in a harness. This results in bulky and large architectures which are cumbersome and restrictive. C-ALEX is a leg exoskeleton without a rigid link structure which has been used in gait training on a treadmill. In this paper, we explore the feasibility of using the C-ALEX exoskeleton over-ground. We converted C-ALEX into a carted system for over-ground use. We tested the architecture on eight healthy subjects to compare the controller's RMS joint torque errors, the effects on step height, joint angles and the deviation of ankle trajectories from target trajectories. The results show that C-ALEX's controller and tension planner have comparable RMS torque errors with no significant difference between the two use cases. C-ALEX is able to increase the step height, affect knee flexion in both walking conditions with no significant difference. There is significant difference between C-ALEX's ability to control hip flexion angles and deviation area over-ground and on a treadmill.

Keywords: Exoskeletons, Design and control, Technology assessment
Abstract: People with Duchenne muscular dystrophy are currently in need of assistive robotics to improve their hand function and have a better quality of life. However, none of the available active hand orthoses is able to address to their specific needs. In this study, the use of hydraulic technology is proposed in the design of an active hand orthosis. Commercially available components were used to identify where customization is necessary for a new electrohydraulic hand orthosis. The presented prototype was able to move four finger modules with a single actuator. The finger modules were separable and had a total mass of only 150 g, whereas the valve manifold added another 250 g. Results revealed that the prototype was able to function well with full flexion/extension cycles up to 2 Hz, but with hysteretic losses between 37-81% of the total input energy. Specialized valves and slave cylinders are required to increase efficiency at higher speeds and to obtain more robust sealing performance.

Keywords: Exoskeletons, Human-machine interfaces, Neuroengineering
Abstract: Robotics rehabilitation is a widely used approach for the treatment of patients with severe motor disabilities, such as stroke survivors. Robots can provide intense, controlled and repeatable rehabilitation and they can also provide different levels of assistance when patients are not able to initiate or complete a movement. Nevertheless, several studies proved that completely passive movements are not sufficient to stimulate neuro-motor recovery and patients’ engagement is a key factor for an effective rehabilitation. For this reason it is important to combine techniques for detection of movement intention (MI) with rehabilitation robotics. In this study we developed an algorithm capable of detecting MI before the movement onset, in order to obtain a trigger signal for providing robotics assistance. The proposed algorithm automatically selects the channels used to extract MI based on the motor-information content of each channel. The developed algorithm was tested on data recorded on n = 8 healthy subjects performing 3D reaching movements with an exoskeleton in active and assisted conditions. MI was detected about 400 ms before the beginning of the movement and the performance of the proposed method were significantly higher than the one achieved when six preselected channels, located over motor areas, were used for MI decoding. MI was also detected during robot-assisted movements. Interestingly, in active movements the highest performance was achieved with electrodes over a well-localized cluster above the contralateral and central motor areas, while in passive executions, the areas with the best performances became more sparse.

Keywords: Wearable and augmenting devices, Design and control, Technology assessment
Abstract: In a previous study, a Tethered Pelvic Assist Device (TPAD) was used to successfully retrain crouch gait of children with Cerebral Palsy by applying a downward force on the pelvis during walking on a treadmill. While the results of this study were promising, an important issue was translating these results to special needs children with crouch gait using simpler alternative procedures. This motivates the present study to compare the biomechanical differences in walking under two conditions: (i) the TPAD applies a pure downward force on the pelvis using tethers, and (ii) a weighted pelvic belt is used to apply the same downward force on the pelvis. In the second case the weight belt also increases the mass at the pelvis. Ten healthy subjects performed two separate experiments while walking on an instrumented treadmill. The whole-body kinematics was recorded using a motion capture system and the ground reaction forces were measured by force plates embedded in the treadmill. We found no significant difference in the actual gait parameters of healthy subjects when the downwards force, equivalent to 15% body weight, applied by the TPAD was replaced by a weighted pelvic belt of 15% body weight. However, the estimated maximum ankle torque, predicted by an inverted pendulum mathematical model, during the single support phase showed a higher increase during walking with the weight belt when compared to a pure downward force. This suggests that the weight belt, due to its simplicity, may be a better medium to translate the results of TPAD in children with cerebral palsy who have a crouch gait.

Keywords: Wearable and augmenting devices, Wearable devices, Exoskeletons
Abstract: Wearable robotic devices and exoskeletons, that assist human beings in physically-demanding tasks have the potential to both increase productivity and reduce the risk of musculoskeletal disorders. Soft exoskeletons, known as exosuit, provide improved portability and fit. While many exosuits are populating the market to support light weights, very few are powerful enough to reinforce workers in lifting heavy loads. Adopting the advantages of novel soft-robotic principles, we propose a voice-controlled upper limb exosuit designed to aid its user in lifting up to 10kg per arm. The exosuit uses a wire-driven mechanism to transmit power from a proximally located actuation stage to the shoulder and elbow. Forces are transmitted to the human body via soft, textile-based components. We evaluate the impact of the device on the muscular effort of a wearer in both a lifting and a holding task. Holding a weight of 14kg with the exosuit results in an average reduction in muscular effort of the biceps brachii and anterior deltoid of 50% and 68%, respectively. Similarly, lifting a weight of 7kg with the exosuit reduces the muscular activity of the same two muscles by 23.4% and 41.2%, respectively.

Keywords: Wearable and augmenting devices, Human-machine interaction
Abstract: Lower back pain is a major health concern worldwide. A cause of lower back pain is the burden on the lumbar region caused by the handling of heavy objects. The Ministry of Health, Labour and Welfare in Japan recommended “squat lifting” to reduce the burden. However, the technique is not very popular although it supports a large force on the lower limbs. Therefore, the aim of this study is to develop a power assist suit for squat lifting. In the study, we propose a knee joint extension mechanism by using a leaf spring. Subsequently, we discuss the estimation of joint torque from the motion analysis of squat lifting to construct a prototype. Finally, we describe the performance of a prototype mounted on a human body. The results indicate that the %MVC of the rectus femoris while performing squat lifting using the prototype is 53% lower than the value obtained without using the prototype.

Keywords: Exoskeletons
Abstract: The load that is carried in military backpacks has drastically increased over the years. Besides energy expenditure, load carriage increases joint loading which has been associated with an increased risk on injuries, discomfort, and reduced task performance. To support soldiers during load carrying, exoskeletons have been proposed. The use of exoskeletons for load carrying is limited since current active exoskeletons have the disadvantage that their power requirements make them unsuitable for long use and (quasi-)passive exoskeletons have mainly focused on metabolic cost. In this paper we present the Exobuddy exoskeleton. The Exobuddy transfers a part of the load directly to the ground. The Exobuddy mechanism is quasi-passive and thereby eliminates the need for large energy sources associated with active exoskeletons. The Exobuddy was evaluated in indoor and outdoor conditions, each completed by four subjects. Exobuddy unloaded the subjects by transferring on average approximately 30% (130 N) of the load to the ground with a maximum of 53% right after heel strike. The energy drawn from the human body to power the quasi-passive mechanism led only to a small, non-significant, increase in energy expenditure. Although not significant, carrying loads with Exobuddy was perceived less exerted and more comfortable compared to carrying loads with the current backpack.

Keywords: Dynamics and control, Biologically-inspired systems, Locomotion and manipulation in robots and biological systems
Abstract: Deploying humanoid robots in complex and unstructured environments requires the development of efficient and adaptive locomotion controllers. Bio-inspiration holds promises in this perspective, since humans are known to have both an energy efficient gait, and the capacity to modulate it across several features like forward speed and step length and height. In this paper, we report the development of a bio-inspired controller for bipedal walking that can achieve controlled modulations of the step height and length over a large range. This controller builds upon our previous work where we combined both a Central Pattern Generator (CPG) and reflex-like modulations with a layer of virtual muscles providing human-like leg impedance. Here, we report first a sensitivity analysis that was performed to identify those among the many parameters of our controller that can actually modulate the step height and length. Then, we report experimental results illustrating such controlled modulations over a large parameter space.

Keywords: Locomotion and manipulation in robots and biological systems, Biomechanics, Dynamics and control
Abstract: This work aims at experimentally identifying a new mechanical descriptor of human locomotion and demonstrating that it can be exploited for the generation of multi-contact motions for humanoids. For this purpose, an experimental setup was built on which five different experiments were carried out by 15 human volunteers. Experimental results show that the distance between the center of mass and the so-called central axis of the external contact wrench significantly varies as a function of locomotion phases and environmental constraints. This finding is combined with a theoretical reasoning in mechanics in order to exhibit how this distance is linked to the whole body's angular acceleration and thus constitutes an interesting parameter to control. Finally, we illustrate the relevance of this result for humanoid robot motion generation by embedding the minimization of the distance between the center of mass and the central axis of the external contact wrench in an optimal control formulation in order to generate multi-contact locomotion in simulation.

Keywords: Human-machine interaction, Wearable and augmenting devices, Design and control
Abstract: Understanding how users adapt their motor behavior to damping forces can improve assistive haptic shared control strategies, for instance in heavy robot-assisted lifting applications. In previous experiments we showed that subjects reaching in constant and position-dependent longitudinal damping fields were able to reduce their movement time and increase end-point accuracy. The movement time versus movement distance and prescribed end-point accuracy agreed with Fitts' Law. However, why subjects were able to have shorter movement time while subjected to impeding damping forces is not explained by Fitts' Law. Based on the minimal variance principle we propose that humans exploit the noise-filtering behavior of constant or position-dependent damping forces. These damping forces attenuate mechanical effects of activation-dependent motor noise. This allows for higher motor activation and shorter movement time without losing end-point accuracy. Consequently, higher allowed motor activation allows for higher accelerations that lead to higher peak velocities, resulting in shorter movement times. Linear and non-linear stochastic optimal feedback control and optimal estimation models with multiplicative noise corroborate measurement data, supporting our hypothesis.

Keywords: Locomotion and manipulation in robots and biological systems, Neuro robotics, Biologically-inspired systems
Abstract: Reproducing human locomotion in simulation has a variety of applications, from informing prosthetic and reha- bilitation medicine to generating stable and human-like robot or animated character movement. In prior work, however, the focus has been on producing stable, natural gaits at a single speed. Novel neuromuscular controllers blending feed- forward and reflex-like control have shown promising success in realizing bio-inspired speed-modulation of walking gaits while adapting a handful of parameters. In this work, we present a modified neuromuscular gait controller in the sagittal plane to similarly realize speed modulation for running gaits. As a result, our controller interpolates fewer than 10 parameters from a stable initialization to realize a large range of running speeds on a simulated bipedal platform. We discuss the speed-evolution and kinematic significance of these selected parameters, and analyze the controller’s velocity-tracking performance over the speed range between 1.3 m/s and 1.7 m/s, which covers much of human running speeds once scaled from platform height.

Keywords: Design and control, Technology assessment, Mobility
Abstract: The sedentary lifestyle of powered wheelchair users has a deleterious effect on their health. If they could exercise while driving their chair, like many manual wheelchair users do, they could potentially improve their health through integrated daily exercise. This paper presents the development of MOVit, a novel, arm exercise-enabling, wheelchair control interface, and the results of three preliminary tests with unimpaired subjects. MOVit consists of two custom-made, instrumented mobile arm supports that replace the armrests of a normal powered wheelchair. Instead of using a joystick to drive the wheelchair, the user moves the arm supports with their arms using a cyclical motion, while the software simulates a “virtual lever drive” chair. MOVit was first tested in a stationary setting with five unimpaired individuals (two expert users and three naïve users) to determine if they could achieve increasing levels of exercise by increasing movement amplitude and frequency. Secondly, driving performance using MOVit was evaluated with the same subjects on a long, straight track. Third, maneuverability with MOVit was evaluated for the expert users. In the stationary setting heart rate and oxygen consumption significantly increased as the level of exercise intensity increased. Driving performance for the long, straight track was comparable to the performance achieved using a standard joystick for the two expert users but was worse for the three novice users due to poor clutching inefficiency. The expert users achieved a level of maneuverability with MOVit comparable to that with a joystick. In conclusion, MOVit can modulate exercise intensity during powered wheelchair driving with a maneuverability comparable to that achieved with a standard joystick.

Keywords: Locomotion and manipulation in robots and biological systems, Biologically-inspired systems, Dynamics and control
Abstract: The emergence and understanding of new design principles that exploit flow-induced mechanical instabilities for propulsion require robust and accurate flow-structure interaction numerical models. In this contribution, we report the development of an algorithm that combines Vortex Particles Mesh (VPM) method and Multi-Body System (MBS) solver for the simulation of actuated swimming structures in fluids. The hydrodynamic efforts are recovered through an innovative approach based on the penalization and projection steps performed within the VPM method. The resulting method avoids time consuming computation of the stresses at the wall to recover the force distribution on the surface of complex deforming shapes. This feature crucially distinguish the proposed approach from other VPM formulations and opens the door for the development of control frameworks for bio-inspired and autonomous robotic swimmers. As a first illustration towards this goal, this paper reports a swimming agent stabilizing its gait in the wake of a cylinder. Illustrating the dynamic features of our framework, we report the energy saved by swimming behind this cylinder as compared to a stationary gait in an induced flow. We also compared this result to the energy saved by following the wake of a moving cylinder.

Keywords: Micro- and nano-robotics
Abstract: In this work, we investigate the near surface effects on the flagellar propulsion of externally actuated soft robotic sperms. A group of 250-μm-long robotic sperms are fabricated using electrospinning, and the influence of a nearby wall on their flagellar propulsion is modeled and characterized inside a fluidic chip with varying width. Our experimental results show that the swimming speed of the robotic sperm decreases by a factor of 2 when its distance to a nearby surface is decreased by 50%, at frequency and precision angle of 5 Hz and 15 degree, respectively. We also show that the reduction in swimming speed can be mitigated by adapting the beating frequency and the precision angle of the tail and head of the robotic sperm during flagellar propulsion. We also demonstrate point-to-point closed-loop control along a reference trajectory inside a fluidic chip with varying width and achieve maximum steady-state error of 5.6 μm.

Keywords: Wearable and augmenting devices, Human-machine interaction, Human-centered design
Abstract: Motor adaptation is a form of motor learning that involves changes in the control of movements that occur as a consequence of optimization in repeated task exposure or practice. A good example of this can be found in human running. It is thought that humans run in accordance with the presence or absence and shape of shoes. Three foot strike patterns exist in running, classified according to which portion of the sole first connects with the ground: RFS (rear-foot-strike; heel connection), MFS (mid-foot-strike, simultaneous heel and toe connections), and FFS (fore-foot-strike; toe, or ball connection). RFS is often seen in runners wearing shoes, whereas FFS is commonly seen during routine barefoot running. This study examined how humans adapt to assistive footwear (environmental exposure)s throughout approximately one month of running training. We found that the shape and stiffness of a modified insole affected the foot, functioning as a spring during running. As training continued, the subject adapted to the structure of the footwear by altering his foot strike pattern while reducing his heart rate. The foot strike pattern changed from heel contact to contact from the side of the toe or arch, and the pattern of the ground reaction force changed from that of RFS to that of FFS or MFS. These results indicate that the change in foot compliance while landing and during grounding affects the foot strike pattern to improve the efficiency and capability of running through long-term training.

Keywords: Exoskeletons, Mobility, Biomechanics
Abstract: The recovery of voluntary arm movements is one of the most important goals during stroke rehabilitation in order to avoid long-term disability in activities of daily living. Support against gravity in order to reduce flexion synergy is reported as an effective strategy to enable upper extremity rehabilitation. In this study, the reduction of muscle activities and muscle forces with the gravity compensated RETRAINER upper limb exoskeleton were analyzed carrying out defined movements with healthy subjects. The EMG signals of the main active muscles as well as the kinematics of different defined motions were captured and compared. The joint kinematics and the joint moments were computationally determined using a 3D musculoskeletal model. The effectiveness of the upper limb gravity compensation could be shown in both mean values of EMG signals and resulting muscle forces, indicating that this compact and lightweight arm exoskeleton can serve as a powerful tool to support the rehabilitation process.

Keywords: Human-machine interaction, Biomechanics, Exoskeletons
Abstract: In this paper, we present an interactive visual task for robot-assisted gait training after stroke. This stand-alone game is interfaced with the impedance controlled modular ankle exoskeleton (“Anklebot”) that provides support only as needed to enhance ankle neuro-motor control in the context of treadmill walking. The interactive task is designed as a simple soccer-based computer video-game such that movement of the game cursor (soccer ball) towards the goal is determined by a patient’s volitional ankle torque. Here, we present the design and features of this interactive video game, as well as the underlying biomechanical model that relates patient-to-game performance. Additionally, we embed simple Statistical analysis algorithms to auto-adjust game parameters in real-time based on patient performance for patient motivation. Finally, we present preliminary test results from a stroke subject trials to validate the video-game performance and its feasibility for clinical use.

Keywords: Neuro robotics, Biomechanics, Neurological disease
Abstract: To what extent gait function will be affected after a stroke depends much on the extent of neurological brain lesion and its location in the brain. Gait in stroke survivors is characterized with certain degree of gait asymmetry, weak or absent forward propulsion on the impaired side, delayed and reduced as well as less coordinated responses to external perturbations. We have recently developed an admittance-controlled Balance Assessment Robot (BAR-TM) that was used in the present research case report in long-term symmetry, push-off and perturbation training with a stroke survivor. Results show that throughout the training selected patient considerably improved gait symmetry, push-off and the timing as well as the response strategy to perturbations.

Keywords: Design and control, Technology assessment, Exoskeletons
Abstract: Robotic gait training is a promising tool for gait rehabilitation in people with neurological disorders. Including intuitive assessment and automatic adaptation of robotic assistance into robotic training is expected to further improve therapy outcomes. This contribution presents a novel performance-based adaptive controller, which adjusts robotic assistance based on the user’s performance for diverse subtasks of gait. The resulting assistance profile of the algorithm could serve as an assessment tool or be used for monitoring progress during therapy. However, during training, values of gait speed and/or partial body weight support (PBWS) might vary. Therefore, the performance criteria should not depend on these factors to result in a reliable assessment. As a first step in deriving the potential of the controller as an assessment tool, ten healthy participants walked in the LOPES II robotic gait trainer testing the adaptive assistance at various gait speeds and levels of PBWS. Performances for all subtasks were dependent on the amount of PBWS. Therefore, the outcome of the novel control algorithm cannot directly be used as an assessment tool, but it has potential to be used for monitoring the progress of patients when the amount of PBWS and the speed are kept constant. Future studies will be focused on further testing the controller on people with neurological disorders to determine its potential as a monitoring tool.

Keywords: Novel actuators, Design and control
Abstract: Abstract— This paper describes the Ambidexter, a low cost portable home-based robotic rehabilitation device for training fine motor skills. The Ambidexter is a 3 degree-of-freedom (DOF) robotic device designed for training hand opening/closing, forearm pronation/supination and wrist flexion/extension. The aim of physical/occupational therapy is to help the patients to improve the ability to perform activities in daily life (ADLs). Currently, due to the high cost and complexity, robotic assisted rehabilitation device are only available at rehabilitation center or therapeutic institution with proper supervision by trained therapist. A low-cost home-based robotic device is needed to solve the existing shortage of trained therapists and high number of patients needing upper limbs rehabilitation. Home-based device also enables patients to get more exercises with minimum assistance at the comfort of their home. It reduces the need to travel and the reliance on physical presence of trained therapists. This paper will present the design considerations and criteria adopted with the aim to reduce cost while maintaining the functionality and effectiveness of the robotic device.

Keywords: Design and control, Human-centered design, Human-machine interaction
Abstract: Arm-weight support has proved to be a key component in robot-aided neuro-rehabilitation in order to permit a wide range of motion for patients with severe disabilities. In this work, a novel motorized platform for sustaining the upper limb of patients during 3D robot-aided rehabilitation and its control architecture are presented. The proposed system is able to support patient’s limb in the 3D space through upper limb kinematic reconstruction during the execution of reaching movements. The platform and the adopted control strategy have been tested on 8 healthy subjects performing point-to-point 3D movements. The trajectory executed by the forearm support has been monitored to assess the performance of the chosen control approach. Moreover, a questionnaire based on the Likert rating scale has been submitted to the subjects to evaluate the overall platform. Preliminary results showed that the proposed control algorithm allowed to follow the arm movement in 3D space with a reduced position error (0.002±0.012 rad). Moreover the subjects felt their arm completely supported, free to move in any direction of the space and judged the platform easy to use.

Keywords: Human-machine interfaces, Design and control, Force control
Abstract: In this paper, we present a haptics-based rehabilitation system that uses kinematics of a haptic device to monitor a subject's participation in therapy. In robot-assisted therapy, it is crucial to monitor if the patient is actively performing the rehabilitation task and is not just passively following the robot's motions. In this paper, we have used position-tracking error patterns as a metric for identifying whether the subject is actively participating in the therapy. Using a single feature identification scheme, our method demonstrated a real-time classification accuracy of 80.04% in separating active and passive participation during a therapy session.

Keywords: Design and control, Human-machine interfaces, Mobility
Abstract: The increased computational complexity demanded by recent algorithms and techniques applied to healthcare and social robotics, often limited by the robot’s embedded hardware, coupled with advancements on networking and cloud computing enabled the so-called cloud robotics paradigm. This work explores cloud robotics concepts pointing at opportunities on the design and development of robotic platforms used for patient mobility assistance. Moreover, we present CloudWalker, a cloud-enabled cyber-physical system to assist mobility impaired individuals. The conception of such system envisions the integration of smart walkers and remote cloud computing platforms, aiming at expanding the range of features these devices can offer to users, patients, healthcare professionals, and family members. Results from validation experiments point to the emergence of a new generation of smart walkers and assistive devices in general, designed to leverage cloud computing concepts to provide an extended range of services to users, relatives, and healthcare professionals.

Keywords: Technology assessment, Wearable devices, Soft robotics
Abstract: Many stroke survivors encounter difficulties in the performance of activities of daily life due to limitations in functional use of the hand. Robotic technology has the potential to compensate for this loss by providing the support that is required to perform activities of daily living, especially when these devices are wearable comfortably for many hours at home. As a first step towards the implementation of assistive technology in the homes of stroke survivors, usability along with the potential effect of prolonged use of a wearable soft-robotic glove during activities of daily life on functional task performance was assessed in this study. Therefore, five chronic stroke survivors were asked to use a wearable soft-robotic glove for four weeks at home during preferred activities of daily life. Before and after the home use of the glove, functional task performance was assessed in a lab environment. After the use of the glove, system usability was assessed. The prolonged use of the glove resulted in an improved supported and unsupported functional performance during tasks related to activities of daily life, as measured with the Jebsen-Taylor Hand Function Test. Promising system usability results were found indicating a good probability for acceptance of the glove. The results from this study indicate the potential of the current glove to be used as assistive tool, which even showed a therapeutic effect. Yet, the glove should be tested in a larger sample for better interpretation and confirmation of these promising results.

Keywords: Technology assessment, Neuroengineering, Human-machine interfaces
Abstract: The paper presents an user study to compare the performance of a tele-rehabilitation robotic system, called HomeRehab, for delivering therapy to stroke patients at home and its version, called PupArm system, designed and developed to provide rehabilitation therapy to patients in clinical settings. Nine patients with different neurological disorders participate in the study. The patients performed 16 movements with each robotic platform and after that they filled a usability survey. Moreover, to evaluate the patient's performance with each robotic device, 8 movement parameters were computed from each trial and for the two robotic devices. Based on the analysis of subjective assessments of usability and the data acquired objectively by the robotic devices, we can conclude that the performance and user experience with both systems are very similar. This finding will be the base of more extensively studies to demonstrate that home-therapy with HomeRehab could be as efficient as therapy in clinical settings assisted by PupArm robot.

Keywords: Technology assessment, Dynamic vision sensors, Human-machine interfaces
Abstract: Autism spectrum disorder (ASD) is a neurodevelopmental disorder that affects people from birth, whose symptoms are found in the early developmental period. The ASD diagnosis is usually performed through several sessions of behavioral observation, exhaustive screening, and manual coding behavior. The early detection of ASD signs in naturalistic behavioral observation may be improved through Child-Robot Interaction (CRI) and technological-based tools for automated behavior assessment. Robot-Assisted Tools using CRI theories have been of interest in intervention for children with Autism Spectrum Disorder (CwASD), elucidating faster and more significant gains from the diagnosis and therapeutic intervention when compared to classical methods. Besides that, using computer vision to analyze child’s behaviors and automated video coding to summarize the responses would help the clinicians to reduce the delay of ASD diagnosis. In this article, a CRI to enhance the traditional tools for ASD diagnosis is proposed, which is inspired on a well-established protocol for Joint Attention assessment, using a computer vision system composed of an unstructured and scalable network of RGBD sensors to allow the child’s movement. Also, a proof of concept is presented, with the participation of three typically developing (TD) children.

Keywords: Neurological disease, Design and control, Human-machine interfaces
Abstract: In this paper we describe the development of an automated therapy device for the use in neurological rehabilitation. The movement patterns of the therapy device belong to a manual therapy method which is based on hippotherapy. This manual therapy method can be used for patients with high movement restrictions because the patient can exercise in a lying position. Those parts of the development of the automated therapy device which we describe in this paper include the movement analysis of the manual therapy method, the mechanic dimensioning, the modeling and the path planning. Furthermore, we illustrate the implementation of our approach into a demonstrator and partially present our validation. The paper serves to demonstrate the clear advantages of developing and using automated assistive therapy devices in neurological rehabilitation.

Keywords: Mobility, Biomechanics, Design and control
Abstract: The ArmAssist, developed by Tecnalia, is a portable cost-effective upper limb rehabilitation platform for at-home tele-rehabilitation after a stroke and a reaching exercise is one of important trainings offered by the ArmAssist. From previous pilot study of the ArmAssist, it has been found that in the reaching exercise, the device should provide comfortable and natural orientation of the forearm for effective and safe rehabilitation. Hence, in this study, we present preliminary measurements of natural orientation of the forearm, specifically yaw angle that corresponds to the orientation of the device during the exercise. Two healthy subjects participated in the measurements using the ArmAssist platform and the results show that comfortable and natural yaw angle of the forearm during the reaching exercise varies with the position while anthropometric information of the subject such as arm length also has an influence on the angle. These findings imply that the forearm position and subject´s limb information should be taken into account to find the proper orientation of the device.

Keywords: Neurological disease, Technology assessment
Abstract: Proprioception – the sense of body awareness – is frequently impaired after stroke. Clinical tests used to detect proprioceptive impairments require substantial amounts of time and lack sensitivity. In contrast, robotic devices demonstrated that they can objectively assess subjects’ proprioceptive acuity. We here present results of a robot-aided bimanual test for wrist proprioception. Ten acute stroke patients with hemiplegic arm weakness and ten healthy subjects participated to the experiment. Subjects actively moved the wrist of the unaffected limb to match the reference position of the affected wrist, which was passively displaced by the robot to a target across six directions. Our results are as follows. First, healthy individuals showed no effect of hand dominance. No acuity differences were found for between dominant or non-dominant hand. Second, the matching error varied significantly across movement directions. Third, subjects with stroke were less accurate than healthy individuals in matching the wrist position. Fourth, the assessment is clinically feasible. Stroke patients completed the test in about 8 minutes. We conclude that assessment of proprioception in stroke should map positions at different joint degrees of freedom to detect underlying proprioceptive abnormalities.

Keywords: Technology assessment, Neurological disease
Abstract: Hand function assessment is essential for upper limb rehabilitation of stroke survivors. Conventional acquisition devices have inherent and restrictive difficulties for their clinical usage. Data gloves are limited for applications outside the medical environment, and motion tracking systems setup are time and personnel demanding. We propose a novel instrument designed as a replica of a glass, equipped with an omnidirectional vision system to capture hand images and an inertial measurement unit for movements kinematic data acquisition. Four stroke survivors were invited as volunteers in pre- and post-treatment experiments for its evaluating. The exercise of drinking water from a glass was elected for the trails. Before treatment, subjects used their contralesional and ipsilateral hands to perform them. Two main functional features were found in the data analysis. There were differences between limbs in the grasping hand postures, mainly in the index and thumb abduction angle, and in the task timing. After treatment, two volunteers repeated the protocol with their contralesional hands. Changes in the features were observed, index and thumb abduction angles were greater in both cases, and tasks timing were altered in distinct ways. These preliminary results suggest the instrument can be used both in evaluation of hand functional deficit and rehabilitation progress. Improvements and future work are also presented.

Keywords: Technology assessment, Mobility
Abstract: As a part of a feasibility analysis, this paper reports that gait analysis of orthopedic-surgically treated patients is possible by using a Kinect v2 sensor as a low cost depth camera instead of applying a conventional marker-based multi-camera system in a gait laboratory. Being aware of the strengths and weaknesses of this approach, our concept expands the potential of gait analysis from diagnostic use only toward the use for documented and actively corrective self-training of patients. The paper analyzes which gait parameters are needed for synthetic, but effective gait evaluation, and if it is possible to obtain them from the Kinect SDK skeleton in terms of exact and stable values compared to parameters obtained by a multi-camera system in a gait laboratory. The main contributions of this paper are the analysis of the usability of the Kinect v2 for mobile gait analysis.

Keywords: Human-machine interfaces, Exoskeletons, Human-machine interaction
Abstract: The aim of this study concerns the evaluation and comparison of different Human-Machine Interfaces for the control of an upper limb motorized exoskeleton for severely impaired patients. Different approaches (i.e. manual, vocal, visual control) are tested in a simulation environment on three subjects affected by muscular dystrophy with the aim of assessing the capability of the system to interact with the user and vice versa. A Graphical User Interface shows the simulated behavior of the exoskeleton to the user which has to perform reaching tasks in the space by moving the exoskeleton end-effector to defined virtual targets that are displayed on the screen. Specific assessment of the interaction of the user with each control interface is achieved, while a quantitative evaluation of the usability of all the three approaches is provided by a System Usability Scale (SUS) questionnaire. All patients were able to interact with all control interfaces without difficulties and to complete reaching tasks in simulation. SUS scores showed overall good usability of the Human-Machine Control Interfaces suggesting that the manual and the vocal control interfaces are preferred by the subjects.

Keywords: Wearable devices, Neuro robotics, Neurological disease
Abstract: The aim of this work is to present a novel wearable mechatronic device (called PDMeter) designed to objectively assess the wrist rigidity in Parkinson's Disease (PD) patients. The system is low-weight, long-term wearable and portable in order to i) perform clinical assessments during Activities of Daily Living (ADLs) in unstructured environments, and ii) to provide several rigidity measurements per day. In this scenario, we defined two different working modalities: i) measurement mode, in which the system measures the wrist rigidity, and ii) backdrivable mode, in which it does not measure, but it has to be transparent for the user during ADLs. In this paper we present the overall mechatronic design of the PDMeter, including the kinematic structure, the actuation system, the sensory system (both force and position) and the control electronics. The overall structure is optimized in terms of dimension and weight: the design of the electronic system allow to integrate in a single compact PCB both the control system and the wireless communication with an external device (laptop or smartphone); the mechanical structure, characterized by one active degree of freedom and five passive ones, is entirely made of aluminium alloy (Al6082), and the whole system with the electronics embedded has an overall mass of about 0.46 kg. Future efforts will focus on the implementation and testing of the most suitable algorithms to assess the wrist rigidity, and their validation in clinical trials.