Search

BackgroundAutism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by persistent deficits in social communication, social interaction, and restricted behaviors. The importance of early intervention has been widely demonstrated, and developmental trajectories in ASD emphasize the importance of nonverbal communication, such as intransitive gesture production, as a possible positive prognostic factor for language development. The use of technological tools in the therapy of individuals with ASD has also become increasingly important due to their higher engagement and responsiveness to technological objects, such as robots.Materials and methodsWe developed a training protocol using the humanoid robot NAO, called IOGIOCO (Interactive mirroring Games wIth sOCial rObot), based on the use of intransitive gestures embedded in naturalistic dialogues, stimulating a triadic interaction between child, robot and therapist. The training was divided into six levels; the first 2 levels were called “familiarization levels,” and the other 4 were “training levels”. The technological setup includes different complexity levels, from mirroring tasks to building spontaneous interactions. We tested the protocol on 10 preschool children with ASD (aged 2–6 years) for 14 weeks. We assessed them at recruitment (T0), at the end of training (T1), and after 6 months (T2).ResultsWe demonstrated the tolerability of the protocol. We found that one group (n=4, males and 2 females) reached the training level, while another and group (n=6 males) remained at a familiarization level (mirroring), we analyzed the results for the two groups. In the group that reached the training levels, we found promising results, such as an improvement in the Social Adaptive Domain of the ABAS-II questionnaire between T0 and T2.ConclusionWhile current results will need a Randomized Controlled Trial to be confirmed, the present work sets an important milestone in using social robots for ASD treatment, aimed at impacting social and communication skills in everyday life.

This work proposes a recurrent neural network-based sim-to-real method to learn mobile robot localization using lidar data in dynamic environments. The main aim of the algorithm is to estimate a Cartesian position error relative to a saved position by means of stored lidar readings in a two-dimensional environment, using lidar data as input. To achieve this, we propose a method that first trains a model on synthetic and augmented LiDAR data to embed rigid transformations into the deep learning model and then fine-tunes the model on real positions using real-world data and external camera measures to produce training labels. This pre-training and fine-tuning approach considerably reduces the time, the computation power, and the amount of real-world data needed to have an accurate model, allowing running the fine-positioning model on the edge of autonomous mobile robots(AMRs). After optimizing the model architecture and hyperparameters, the devised model is tested in different scenarios, comparing the precise positioning capability of AMRs with that of a classical iterative closest point and advanced Monte Carlo localization.

In modern applied research concerning human-machine interaction, the possibility of increasing the degree of intelligence and dexterity of the controlled plant by imitating human control dynamics has been a primary objective. To reach this goal, we propose a novel nonlinear modeling technique able to predict human force generated during a cooperative task with a controlled robot. The proposed Narmax model was constructed using an artificial neural network as a nonlinear functional approximator and was firstly trained offline with data acquired from ten subjects, performing a manipulation task on a small collaborative robot. Then, given the complexity of the system and the characteristics of human response, the exploitation of Peak to Peak Dynamics allowed the development of a reduced-order model that could reliably forecast the peak of human response. Ultimately, our human model was tested online on an industrial high-payload robot, showing its general applicability and how it can be used to let the robot anticipate human intention during collaborative manipulation.

In the context of neurorehabilitation, there have been rapid and continuous improvements in sensors-based clinical tools to quantify limb performance. As a result of the increasing integration of technologies in the assessment procedure, the need to integrate evidence-based medicine with benchmarking has emerged in the scientific community. In this work, we present the experimental validation of our previously proposed benchmarking scheme for upper limb capabilities in terms of repeatability, reproducibility, and clinical meaningfulness. We performed a prospective multicenter study on neurologically intact young and elderly subjects and post-stroke patients while recording kinematics and electromyography. 60 subjects (30 young healthy, 15 elderly healthy, and 15 post-stroke) completed the benchmarking protocol. The framework was repeatable among different assessors and instrumentation. Age did not significantly impact the performance indicators of the scheme for healthy subjects. In post-stroke subjects, the movements presented decreased smoothness and speed, the movement amplitude was reduced, and the muscular activation showed lower power and lower intra-limb coordination. We revised the original framework reducing it to three motor skills, and we extracted 14 significant performance indicators with a good correlation with the ARAT clinical scale. The applicability of the scheme is wide, and it may be considered a valuable tool for upper limb functional evaluation in the clinical routine.

Children with Autism Spectrum Disorder (ASD) show severe attention deficits, hindering their capacity to acquire new skills. The automatic assessment of their attention response would provide the therapists with an important biomarker to better quantify their behaviour and monitor their progress during therapy. This work aims to develop a quantitative model, to evaluate the attention response of children with ASD, during robotic-assistive therapeutic sessions. Previous attempts to quantify the attention response of autistic subjects during human-robot interaction tasks were limited to restrained child movements. Instead, we developed an accurate quantitative system to assess the attention of ASD children in unconstrained scenarios. Our approach combines gaze extraction (Gaze360 model) with the definition of angular Areas-of-Interest, to characterise periods of attention towards elements of interest in the therapy environment during the session. The methodology was tested with 12 ASD children, achieving a mean test accuracy of 79.5 %. Finally, the proposed attention index was consistent with the therapists’ evaluation of patients, allowing a meaningful interpretation of the automatic evaluation. This encourages the future clinical use of the proposed system.

Abstract Introduction To date, lung cancer is one of the most lethal diagnoses worldwide. A variety of lung cancer treatments and modalities are available, which are generally presented during the patient and doctor consultation. The implementation of decision tools to facilitate patient's decision‐making and the management of their healthcare process during medical consultation is fundamental. Studies have demonstrated that decision tools are helpful to promote health management and decision‐making of lung cancer patients during consultations. The main aim of the present work within the I3LUNG project is to systematically review the implementation of decision tools to facilitate medical consultation about oncological treatments for lung cancer patients. Methods In the present study, we conducted a systematic review following the PRISMA guidelines. We used an electronic computer‐based search involving three databases, as follows: Embase, PubMed, and Scopus. 10 articles met the inclusion criteria and were included. They explicitly refer to decision tools in the oncological context, with lung cancer patients. Results The discussion highlights the most encouraging results about the positive role of decision aids during medical consultations about oncological treatments, especially regarding anxiety, decision‐making, and patient knowledge. However, no one main decision aid tool emerged as essential. Opting for a more recent timeframe to select eligible articles might shed light on the current array of decision aid tools available. Conclusion Future review efforts could utilize alternative search strategies to explore other lung cancer‐specific outcomes during medical consultations for treatment decisions and the implementation of decision aid tools. Engaging with experts in the fields of oncology, patient decision‐making, or health communication could provide valuable insights and recommendations for relevant literature or research directions that may not be readily accessible through traditional search methods. The development of guidelines for future research were provided with the aim to promote decision aids focused on patients' needs.

According to the motor learning theory by Albus and Ito, synaptic depression at the parallel fibre to Purkinje cells synapse (pf-PC) is the main substrate responsible for learning sensorimotor contingencies under climbing fibre control. However, recent experimental evidence challenges this relatively monopolistic view of cerebellar learning. Bidirectional plasticity appears crucial for learning, in which different microzones can undergo opposite changes of synaptic strength (e.g. downbound microzones-more likely depression, upbound microzones-more likely potentiation), and multiple forms of plasticity have been identified, distributed over different cerebellar circuit synapses. Here, we have simulated classical eyeblink conditioning (CEBC) using an advanced spiking cerebellar model embedding downbound and upbound modules that are subject to multiple plasticity rules. Simulations indicate that synaptic plasticity regulates the cascade of precise spiking patterns spreading throughout the cerebellar cortex and cerebellar nuclei. CEBC was supported by plasticity at the pf-PC synapses as well as at the synapses of the molecular layer interneurons (MLIs), but only the combined switch-off of both sites of plasticity compromised learning significantly. By differentially engaging climbing fibre information and related forms of synaptic plasticity, both microzones contributed to generate a well-timed conditioned response, but it was the downbound module that played the major role in this process. The outcomes of our simulations closely align with the behavioural and electrophysiological phenotypes of mutant mice suffering from cell-specific mutations that affect processing of their PC and/or MLI synapses. Our data highlight that a synergy of bidirectional plasticity rules distributed across the cerebellum can facilitate finetuning of adaptive associative behaviours at a high spatiotemporal resolution.

Abstract Human-robot interaction is a rapidly developing field and robots have been taking more active roles in our daily lives. Patient care is one of the fields in which robots are becoming more present, especially for people with disabilities. People with neurodegenerative disorders might not consciously or voluntarily produce movements other than those involving the eyes or eyelids. In this context, Brain-Computer Interface (BCI) systems present an alternative way to communicate or interact with the external world. In order to improve the lives of people with disabilities, this paper presents a novel BCI to control an assistive robot with user’s eye artifacts. In this study, eye artifacts that contaminate the electroencephalogram (EEG) signals are considered a valuable source of information thanks to their high signal-to-noise ratio and intentional generation. The proposed methodology detects eye artifacts from EEG signals through characteristic shapes that occur during the events. The lateral movements are distinguished by their ordered peak and valley formation and the opposite phase of the signals measured at F7 and F8 channels. This work, as far as the authors’ knowledge, is the first method that used this behavior to detect lateral eye movements. For the blinks detection, a double-thresholding method is proposed by the authors to catch both weak blinks as well as regular ones, differentiating itself from the other algorithms in the literature that normally use only one threshold. Real-time detected events with their virtual time stamps are fed into a second algorithm, to further distinguish between double and quadruple blinks from single blinks occurrence frequency. After testing the algorithm offline and in realtime, the algorithm is implemented on the device. The created BCI was used to control an assistive robot through a graphical user interface. The validation experiments including 5 participants prove that the developed BCI is able to control the robot.

Abstract Background Robot-generated deviating forces during multijoint reaching movements have been applied to investigate motor control and to tune neuromotor adaptation. Can the application of force to limbs improve motor learning? In this framework, the response to altered dynamic environments of children affected by primary dystonia has never been studied. Methods As preliminary pilot study, eleven children with primary dystonia and eleven age-matched healthy control subjects were asked to perform upper limb movements, triangle-reaching (three directions) and circle-writing, using a haptic robot interacting with ad-hoc developed task-specific visual interfaces. Three dynamic conditions were provided, null additive external force (A), constant disturbing force (B) and deactivation of the additive external force again (C). The path length for each trial was computed, from the recorded position data and interaction events. Results The results show that the disturbing force affects significantly the movement outcomes in healthy but not in dystonic subjects, already compromised in the reference condition: the external alteration uncalibrates the healthy sensorimotor system, while the dystonic one is already strongly uncalibrated. The lack of systematic compensation for perturbation effects during B condition is reflected into the absence of after-effects in C condition, which would be the evidence that CNS generates a prediction of the perturbing forces using an internal model of the environment. The most promising finding is that in dystonic population the altered dynamic exposure seems to induce a subsequent improvement, i.e. a beneficial after-effect in terms of optimal path control, compared with the correspondent reference movement outcome. Conclusions The short-time error-enhancing training in dystonia could represent an effective approach for motor performance improvement, since the exposure to controlled dynamic alterations induces a refining of the existing but strongly imprecise motor scheme and sensorimotor patterns.

Abstract Background The restoration of walking ability is the main goal of post-stroke lower limb rehabilitation and different studies suggest that pedaling may have a positive effect on locomotion. The aim of this study was to explore the feasibility of a biofeedback pedaling treatment and its effects on cycling and walking ability in chronic stroke patients. A case series study was designed and participants were recruited based on a gait pattern classification of a population of 153 chronic stroke patients. Methods In order to optimize participants selection, a k-means cluster analysis was performed to subgroup homogenous gait patterns in terms of gait speed and symmetry. The training consisted of a 2-week treatment of 6 sessions. A visual biofeedback helped the subjects in maintaining a symmetrical contribution of the two legs during pedaling. Participants were assessed before, after training and at follow-up visits (one week after treatment). Outcome measures were the unbalance during a pedaling test, and the temporal, spatial, and symmetry parameters during gait analysis. Results and discussion Three clusters, mainly differing in terms of gait speed, were identified and participants, representative of each cluster, were selected. An intra-subject statistical analysis (ANOVA) showed that all patients significantly decreased the pedaling unbalance after treatment and maintained significant improvements with respect to baseline at follow-up. The 2-week treatment induced some modifications in the gait pattern of two patients: one, the most impaired, significantly improved mean velocity and increased gait symmetry; the other one reduced significantly the over-compensation of the healthy limb. No benefits were produced in the gait of the last subject who maintained her slow but almost symmetrical pattern. Thus, this study might suggest that the treatment can be beneficial for patients having a very asymmetrical and inefficient gait and for those that overuse the healthy leg. Conclusion The results demonstrated that the treatment is feasible and it might be effective in translating progresses from pedaling to locomotion. If these results are confirmed on a larger and controlled scale, the intervention, thanks to its safety and low price, could have a significant impact as a home- rehabilitation treatment for chronic stroke patients.

Abstract Background Correlating the features of the actual executed movement with the associated cortical activations can enhance the reliability of the functional Magnetic Resonance Imaging (fMRI) data interpretation. This is crucial for longitudinal evaluation of motor recovery in neurological patients and for investigating detailed mutual interactions between activation maps and movement parameters. Therefore, we have explored a new set-up combining fMRI with an optoelectronic motion capture system, which provides a multi-parameter quantification of the performed motor task. Methods The cameras of the motion system were mounted inside the MR room and passive markers were placed on the subject skin, without any risk or encumbrance. The versatile set-up allows 3-dimensional multi-segment acquisitions including recording of possible mirror movements, and it guarantees a high inter-sessions repeatability. We demonstrated the integrated set-up reliability through compatibility tests. Then, an fMRI block-design protocol combined with kinematic recordings was tested on a healthy volunteer performing finger tapping and ankle dorsal- plantar-flexion. A preliminary assessment of clinical applicability and perspectives was carried out by pre- and post rehabilitation acquisitions on a hemiparetic patient performing ankle dorsal- plantar-flexion. For all sessions, the proposed method integrating kinematic data into the model design was compared with the standard analysis. Results Phantom acquisitions demonstrated the not-compromised image quality. Healthy subject sessions showed the protocols feasibility and the model reliability with the kinematic regressor. The patient results showed that brain activation maps were more consistent when the images analysis included in the regression model, besides the stimuli, the kinematic regressor quantifying the actual executed movement (movement timing and amplitude), proving a significant model improvement. Moreover, concerning motor recovery evaluation, after one rehabilitation month, a greater cortical area was activated during exercise, in contrast to the usual focalization associated with functional recovery. Indeed, the availability of kinematics data allows to correlate this wider area with a higher frequency and a larger amplitude of movement. Conclusions The kinematic acquisitions resulted to be reliable and versatile to enrich the fMRI images information and therefore the evaluation of motor recovery in neurological patients where large differences between required and performed motion can be expected.

Abstract Background The design of an optimal neuroprostheses controller and its clinical use presents several challenges. First, the physiological system is characterized by highly inter-subjects varying properties and also by non stationary behaviour with time, due to conditioning level and fatigue. Secondly, the easiness to use in routine clinical practice requires experienced operators. Therefore, feedback controllers, avoiding long setting procedures, are required. Methods The error mapping controller (EMC) here proposed uses artificial neural networks (ANNs) both for the design of an inverse model and of a feedback controller. A neuromuscular model is used to validate the performance of the controllers in simulations. The EMC performance is compared to a Proportional Integral Derivative (PID) included in an anti wind-up scheme (called PIDAW) and to a controller with an ANN as inverse model and a PID in the feedback loop (NEUROPID). In addition tests on the EMC robustness in response to variations of the Plant parameters and to mechanical disturbances are carried out. Results The EMC shows improvements with respect to the other controllers in tracking accuracy, capability to prolong exercise managing fatigue, robustness to parameter variations and resistance to mechanical disturbances. Conclusion Different from the other controllers, the EMC is capable of balancing between tracking accuracy and mapping of fatigue during the exercise. In this way, it avoids overstressing muscles and allows a considerable prolongation of the movement. The collection of the training sets does not require any particular experimental setting and can be introduced in routine clinical practice.

Objective: The aim of the present study is to explore whether a single session of transcutaneous Vagus Nerve Stimulation (tVNS) can enhance the ipsilesional, and contralesional upper limb motor functions as well as cognitive functions in stroke patients. The effects of the stimulation were evaluated through two different tasks: the box and blocks test (BB), indexing manual dexterity, and the Go/No-go task, a visuomotor paradigm used to assess both motor readiness and response inhibition. Tests were administered without tVNS, during tVNS and during sham tVNS. Results: The BB showed a statistical difference for both contralesional side (p = 0.05) between Basal-Real condition (p = 0.042) and ipsilesional side (p = 0.001) between Basal-Real (p = 0.008) and for Real-Sham (p = 0.005). Any statistical difference was found for the mean latencies in the three conditions of the Go/No-go test. Conclusion: A single session of tVNS seems to improve upper limb motor functions but not cognitive functions in post-stroke patients, despite a positive trend was detected.

In Human–Machine interaction, the possibility of increasing the intelligence and adaptability of the controlled plant by imitating human control behavior has been an objective of many research efforts in the last decades. From classical control-theory human control models to modern machine learning, neural networks, and reinforcement learning paradigms, the common denominator is the effort to model complex nonlinear dynamics typical of human activity. However, these approaches are very different, and finding a guiding line is challenging. This review investigates state-of-the-art techniques from the perspective of human control modeling, considering the different physiological districts involved as the starting point. The focus is mainly directed toward nonlinear dynamical system modeling, which constitutes the main challenge in this field. In the end, transport systems are presented as a technological scenario in which the discussed techniques are mainly applied.

Abstract Nowadays, work-related musculoskeletal disorders have a drastic impact on a large part of the world population. In particular, low-back pain counts as the leading cause of absence from work in the industrial sector. Robotic exoskeletons have great potential to improve industrial workers’ health and life quality. Nonetheless, current solutions are often limited by sub-optimal control systems. Due to the dynamic environment in which they are used, failure to adapt to the wearer and the task may be limiting exoskeleton adoption in occupational scenarios. In this scope, we present a deep-learning-based approach exploiting inertial sensors to provide industrial exoskeletons with human activity recognition and adaptive payload compensation. Inertial measurement units are easily wearable or embeddable in any industrial exoskeleton. We exploited Long-Short Term Memory networks both to perform human activity recognition and to classify the weight of lifted objects up to 15 kg. We found a median F1 score of $$90.80\%$$ 90.80 % (activity recognition) and $$87.14\%$$ 87.14 % (payload estimation) with subject-specific models trained and tested on 12 (6M-6F) young healthy volunteers. We also succeeded in evaluating the applicability of this approach with an in-lab real-time test in a simulated target scenario. These high-level algorithms may be useful to fully exploit the potential of powered exoskeletons to achieve symbiotic human–robot interaction.

In many real-world applications (e.g., human-robot collaboration), the environment changes rapidly, and the intended path may become invalid due to moving obstacles. In these situations, the robot should quickly find a new path to reach the goal, possibly without stopping. Planning from scratch or repairing the current graph can be too expensive and time-consuming. This paper proposes MARS, a sampling-based Multi-pAth Replanning Strategy that enables a robot to move in dynamic environments with unpredictable obstacles. The novelty of the method is the exploitation of a set of precomputed paths to compute a new solution in a few hundred milliseconds when an obstacle obstructs the robot’s path. The algorithm enhances the search speed by using informed sampling, builds a directed graph to reuse results from previous replanning iterations, and improves the current solution in an anytime fashion to make the robot responsive to environmental changes. In addition, the paper presents a multithread architecture, applicable to several replanning algorithms, to handle the execution of the robot’s trajectory with continuous replanning and the collision checking of the traversed path. The paper compares state-of-the-art sampling-based path-replanning algorithms in complex and high-dimensional scenarios, showing that MARS is superior in terms of success rate and quality of solutions found. An open-source ROS-compatible implementation of the algorithm is also provided.