Physically interacting humans regulate muscle coactivation to improve visuo-haptic perception.

When moving a piano or dancing tango with a partner, how should I control my arm muscles to sense their movements and follow or guide them smoothly? Here we observe how physically connected pairs tracking a moving target with the arm modify muscle coactivation with their visual acuity and the partner's performance. They coactivate muscles to stiffen the arm when the partner's performance is worse and relax with blurry visual feedback. Computational modeling shows that this adaptive sensing property cannot be explained by the minimization of movement error hypothesis that has previously explained adaptation in dynamic environments. Instead, individuals skillfully control the stiffness to guide the arm toward the planned motion while minimizing effort and extracting useful information from the partner's movement. The central nervous system regulates muscle activation to guide motion with accurate task information from vision and haptics while minimizing the metabolic cost. As a consequence, the partner with the most accurate target information leads the movement.NEW & NOTEWORTHY Our results reveal that interacting humans inconspicuously modulate muscle activation to extract accurate information about the common target while considering their own and the partner's sensorimotor noise. A novel computational model was developed to decipher the underlying mechanism: muscle coactivation is adapted to combine haptic information from the interaction with the partner and own visual information in a stochastically optimal manner. This improves the prediction of the target position with minimal metabolic cost in each partner, resulting in the lead of the partner with the most accurate visual information.

Uncertainty-aware automated assessment of the arm impedance with upper-limb exoskeletons.

Providing high degree of personalization to a specific need of each patient is invaluable to improve the utility of robot-driven neurorehabilitation. For the desired customization of treatment strategies, precise and reliable estimation of the patient's state becomes important, as it can be used to continuously monitor the patient during training and to document the rehabilitation progress. Wearable robotics have emerged as a valuable tool for this quantitative assessment as the actuation and sensing are performed on the joint level. However, upper-limb exoskeletons introduce various sources of uncertainty, which primarily result from the complex interaction dynamics at the physical interface between the patient and the robotic device. These sources of uncertainty must be considered to ensure the correctness of estimation results when performing the clinical assessment of the patient state. In this work, we analyze these sources of uncertainty and quantify their influence on the estimation of the human arm impedance. We argue that this mitigates the risk of relying on overconfident estimates and promotes more precise computational approaches in robot-based neurorehabilitation.

Model-Based Shared Control of a Hybrid FES-Exoskeleton: An Application in Participant-Specific Robotic Rehabilitation.

Hybrid exoskeleton, comprising an exoskeleton interfaced with functional electrical stimulation (FES) technique, is conceptualized to complement the weakness of each other in automated neuro-rehabilitation of sensory-motor deficits. The externally actuating exoskeleton cannot directly influence neurophysiology of the patients, while FES is difficult to use in functional or goal-oriented tasks. The latter challenge is largely inherited from the fact that the dynamics of the muscular response to FES is complex, and it is highly user- and state-dependent. Due to the retardation of the muscular contraction response to the FES profile, furthermore, a commonly used model-free control scheme, such as PID control, suffers performance. The challenge in FES control is exacerbated especially in the presence of the actuation redundancy between the volitional activity of the user, powered exoskeleton, and FES-induced muscle contractions. This study therefore presents trajectory tracking performance of the hybrid exoskeleton in a novel model-based hybrid exoskeleton scheme which entices user-specific FES model-predictive control.

Assessing Human-Human Kinematics for the Implementation of Robot-Assisted Physical Therapy in Humanoids: A Pilot Study.

The development of humanoids with bimanual manipulator arms may facilitate assistive robots to perform physical therapy with older adults living at home. As we assume the human-human interaction to be the gold standard of physical therapy, we propose a kinematics analysis to derive guidelines for implementing physical therapy assisted by humanoids. Therefore, a pilot study was carried out involving three physical therapists and two participants acting as exemplary patients. The study analyzes the therapists' movement strategy, including the position and orientation of the therapists' bodies in relation to the participants and the placement of the therapists' hands on the upper limb segment of the participants, as well as the inter- and intravariability during the performance of a ROM (range of motion) assessment. The results demonstrate that while physical therapists exhibit variation in their interaction strategies, they still achieve a consistently low level of variability in their manipulation space.

Hybrid Robotic and Electrical Stimulation Assistance Can Enhance Performance and Reduce Mental Demand.

Combining functional electrical stimulation (FES) and robotics may enhance recovery after stroke, by providing neural feedback with the former while improving quality of motion and minimizing muscular fatigue with the latter. Here, we explored whether and how FES, robot assistance and their combination, affect users' performance, effort, fatigue and user experience. 15 healthy participants performed a wrist flexion/extension tracking task with FES and/or robotic assistance. Tracking performance improved during the hybrid FES-robot and the robot-only assistance conditions in comparison to no assistance, but no improvement is observed when only FES is used. Fatigue, muscular and voluntary effort are estimated from electromyographic recording. Total muscle contraction and volitional activity are lowest with robotic assistance, whereas fatigue level do not change between the conditions. The NASA-Task Load Index answers indicate that participants found the task less mentally demanding during the hybrid and robot conditions than the FES condition. The addition of robotic assistance to FES training might thus facilitate an increased user engagement compared to robot training and allow longer motor training session than with FES assistance.

Online detection of compensatory strategies in human movement with supervised classification: a pilot study.

Introduction: Stroke survivors often compensate for the loss of motor function in their distal joints by altered use of more proximal joints and body segments. Since this can be detrimental to the rehabilitation process in the long-term, it is imperative that such movements are indicated to the patients and their caregiver. This is a difficult task since compensation strategies are varied and multi-faceted. Recent works that have focused on supervised machine learning methods for compensation detection often require a large training dataset of motions with compensation location annotations for each time-step of the recorded motion. In contrast, this study proposed a novel approach that learned a linear classifier from energy-based features to discriminate between healthy and compensatory movements and identify the compensating joints without the need for dense and explicit annotations. Methods: Six healthy physiotherapists performed five different tasks using healthy movements and acted compensations. The resulting motion capture data was transformed into joint kinematic and dynamic trajectories. Inspired by works in bio-mechanics, energy-based features were extracted from this dataset. Support vector machine (SVM) and logistic regression (LR) algorithms were then applied for detection of compensatory movements. For compensating joint identification, an additional condition enforcing the independence of the feature calculation for each observable degree of freedom was imposed. Results: Using leave-one-out cross validation, low values of mean brier score (<0.15), mis-classification rate (<0.2) and false discovery rate (<0.2) were obtained for both SVM and LR classifiers. These methods were found to outperform deep learning classifiers that did not use energy-based features. Additionally, online classification performance by our methods were also shown to outperform deep learning baselines. Furthermore, qualitative results obtained from the compensation joint identification experiment indicated that the method could successfully identify compensating joints. Discussion: Results from this study indicated that including prior bio-mechanical information in the form of energy based features can improve classification performance even when linear classifiers are used, both for offline and online classification. Furthermore, evaluation compensation joint identification algorithm indicated that it could potentially provide a straightforward and interpretable way of identifying compensating joints, as well as the degree of compensation being performed.

Hybrid Functional Electrical Stimulation and Robotic Assistance for Wrist Motion Training After Stroke: Preliminary Results.

This work presents preliminary results of a clinical study with sub-acute stroke patients using a hybrid system for wrist rehabilitation. The patients trained their wrist flexion/extension motion through a target tracking task, where electrical stimulation and robotic torque assisted them proportionally to their tracking error. Five sub-acute stroke patients have completed the training for 3 sessions on separate days. The preliminary results show hybrid assistance improves tracking performance and motion smoothness in most participants. In each session, patients' tracking performances before and after training were evaluated in unassisted tracking trials, without assistance. Their unassisted performance was compared across sessions and the results suggest that moderately to severely impaired patients might benefit more from hybrid training with our system than mildly impaired patients. Subjective assessments from all sessions show that the patients found the use of the device very comfortable and the training enjoyable. More data is being collected and future work will aim at verifying these trends.

Modeling inter-human movement coordination: synchronization governs joint task dynamics.

Human interaction partners tend to synchronize their movements during repetitive actions such as walking. Research of inter-human coordination in purely rhythmic action tasks reveals that the observed patterns of interaction are dominated by synchronization effects. Initiated by our finding that human dyads synchronize their arm movements even in a goal-directed action task, we present a step-wise approach to a model of inter-human movement coordination. In an experiment, the hand trajectories of ten human dyads are recorded. Governed by a dynamical process of phase synchronization, the participants establish in-phase as well as anti-phase relations. The emerging relations are successfully reproduced by the attractor dynamics of coupled phase oscillators inspired by the Kuramoto model. Three different methods on transforming the motion trajectories into instantaneous phases are investigated and their influence on the model fit to the experimental data is evaluated. System identification technique allows us to estimate the model parameters, which are the coupling strength and the frequency detuning among the dyad. The stability properties of the identified model match the relations observed in the experimental data. In short, our model predicts the dynamics of inter-human movement coordination. It can directly be implemented to enrich human-robot interaction.

Estimation of Involuntary Components of Human Arm Impedance in Multi-Joint Movements via Feedback Jerk Isolation.

Stable and efficient coordination in physical human-robot interaction requires consideration of human feedback behavior. In unpredictable tasks, where voluntary cognitive feedback is too slow to guarantee desired task execution, the human must rely on involuntary intrinsic and reflexive feedback. The combined effects of these two feedback mechanisms and the inertial characteristics can be summarized in the involuntary impedance components. In this work, we present a method for the estimation of the involuntary impedance components of the human arm in multi-joint movements. We apply force perturbations to evoke feedback jerks that can be isolated using a high pass filter and limit the duration of the estimation interval to guarantee exclusion of voluntary cognitive feedback. Dynamic regressor representation of the rigid body dynamics of the arm and first order Taylor series expansion of the feedback behavior yield a model that is linear in the involuntary impedance components. The constant values of the inertial parameters are estimated in a static posture maintenance task and subsequently inserted to estimate the remaining components in a dynamic movement task. The method is validated with simulated data of a neuromechanical model of the human arm and its performance is compared to established methods from the literature. The results of the validation demonstrate superior estimation performance for moderate movement velocities, and less influence of the variability of the movements. The applicability to real data and the plausibility of the limited estimation interval are successfully demonstrated in an experiment with human participants.

Effect of External Force on Agency in Physical Human-Machine Interaction.

In the advent of intelligent robotic tools for physically assisting humans, user experience, and intuitiveness in particular have become important features for control designs. However, existing works predominantly focus on performance-related measures for evaluating control systems as the subjective experience of a user by large cannot be directly observed. In this study, we therefore focus on agency-related interactions between control and embodiment in the context of physical human-machine interaction. By applying an intentional binding paradigm in a virtual, machine-assisted reaching task, we evaluate how the sense of agency of able-bodied humans is modulated by assistive force characteristics of a physically coupled device. In addition to measuring how assistive force profiles influence the sense of agency with intentional binding, we analyzed the sense of agency using a questionnaire. Remarkably, our participants reported to experience stronger agency when being appropriately assisted, although they contributed less to the control task. This is substantiated by the overall consistency of intentional binding results and the self-reported sense of agency. Our results confirm the fundamental feasibility of the sense of agency to objectively evaluate the quality of human-in-the-loop control for assistive technologies. While the underlying mechanisms causing the perceptual bias observed in the intentional binding paradigm are still to be understood, we believe that this study distinctly contributes to demonstrating how the sense of agency characterizes intuitiveness of assistance in physical human-machine interaction.

High-Resolution Motor State Detection in Parkinson's Disease Using Convolutional Neural Networks.

Patients with advanced Parkinson's disease regularly experience unstable motor states. Objective and reliable monitoring of these fluctuations is an unmet need. We used deep learning to classify motion data from a single wrist-worn IMU sensor recording in unscripted environments. For validation purposes, patients were accompanied by a movement disorder expert, and their motor state was passively evaluated every minute. We acquired a dataset of 8,661 minutes of IMU data from 30 patients, with annotations about the motor state (OFF,ON, DYSKINETIC) based on MDS-UPDRS global bradykinesia item and the AIMS upper limb dyskinesia item. Using a 1-minute window size as an input for a convolutional neural network trained on data from a subset of patients, we achieved a three-class balanced accuracy of 0.654 on data from previously unseen subjects. This corresponds to detecting the OFF, ON, or DYSKINETIC motor state at a sensitivity/specificity of 0.64/0.89, 0.67/0.67 and 0.64/0.89, respectively. On average, the model outputs were highly correlated with the annotation on a per subject scale (r = 0.83/0.84; p < 0.0001), and sustained so for the highly resolved time windows of 1 minute (r = 0.64/0.70; p < 0.0001). Thus, we demonstrate the feasibility of long-term motor-state detection in a free-living setting with deep learning using motion data from a single IMU.

Impedance-Based Gaussian Processes for Modeling Human Motor Behavior in Physical and Non-Physical Interaction.

OBJECTIVE: Modeling of human motor intention plays an essential role in predictively controlling a robotic system in human-robot interaction tasks. In most machine learning techniques, human motor behavior is modeled as a generic stochastic process. However, the integration of a priori knowledge about underlying system structures can provide insights on otherwise unobservable intrinsic states that yield the superior prediction performance and increased generalization capabilities. METHODS: We present a novel method for modeling human motor behavior that explicitly includes a neuroscientifically supported model of human motor control, in which the dynamics of the human arm are modeled by a mechanical impedance that tracks a latent desired trajectory. We adopt a Bayesian setting by defining Gaussian process (GP) priors for the impedance elements and the latent desired trajectory. This enables exploitation of a priori human arm impedance knowledge for regression of interaction forces through inference of a latent desired human trajectory. RESULTS: The method is validated using simulated data, with particular focus on effects of GP prior parameterization and intention estimation capabilities. The superior prediction performance is shown with respect to a naive GP prior. An experiment with human participants evaluates generalization capabilities and effects of training data sparsity. CONCLUSION: We derive the correlations of an impedance-based GP model of human motor behavior that exploits a priori knowledge. SIGNIFICANCE: The model effectively predicts interaction forces by inferring a latent desired human trajectory in previously observed as well as unobserved regions of the input space.

Human-Robot Team Interaction Through Wearable Haptics for Cooperative Manipulation.

The interaction of robot teams and single human in teleoperation scenarios is beneficial in cooperative tasks, for example, the manipulation of heavy and large objects in remote or dangerous environments. The main control challenge of the interaction is its asymmetry, arising because robot teams have a relatively high number of controllable degrees of freedom compared to the human operator. Therefore, we propose a control scheme that establishes the interaction on spaces of reduced dimensionality taking into account the low number of human command and feedback signals imposed by haptic devices. We evaluate the suitability of wearable haptic fingertip devices for multi-contact teleoperation in a user study. The results show that the proposed control approach is appropriate for human-robot team interaction and that the wearable haptic fingertip devices provide suitable assistance in cooperative manipulation tasks.

The effect of skill level matching in dyadic interaction on learning of a tracing task.

Dyadic interaction between humans has gained great research interest in the last years. The effects of factors that influence the interaction, as e.g. roles or skill level matching, are still not well understood. In this paper, we further investigated the effect of skill level matching between partners on learning of a visuo-motor task. Understanding the effect of skill level matching is crucial for applications in collaborative rehabilitation. Fifteen healthy participants were asked to trace a path while being subjected to a visuo-motor rotation (Novice). The Novices were paired with a partner, forming one of the three Dyad Types: a) haptic connection to another Novice, b) haptic connection to an Expert (no visuo-motor rotation), or c) no haptic. The intervention consisted of a Familiarization phase, followed by a Training phase, in which the Novices were learning the task in the respective Dyad Type, and a Test phase in which the learning was assessed (haptic connection removed, if any). Results suggest that learning of the task with a haptic connection to an Expert was least beneficial. However, during the Training phase the dyads comprising an Expert clearly outperformed the dyads with matched skill levels. The results point towards the same direction as previous findings in literature and can be explained by current motor-learning theories. Future work needs to corroborate these preliminary results.

A Multi-Layer Gaussian Process for Motor Symptom Estimation in People With Parkinson's Disease.

The assessment of Parkinson's disease (PD) poses a significant challenge, as it is influenced by various factors that lead to a complex and fluctuating symptom manifestation. Thus, a frequent and objective PD assessment is highly valuable for effective health management of people with Parkinson's disease (PwP). Here, we propose a method for monitoring PwP by stochastically modeling the relationships between wrist movements during unscripted daily activities and corresponding annotations about clinical displays of movement abnormalities. We approach the estimation of PD motor signs by independently modeling and hierarchically stacking Gaussian process models for three classes of commonly observed movement abnormalities in PwP including tremor, (non-tremulous) bradykinesia, and (non-tremulous) dyskinesia. We use clinically adopted severity measures as annotations for training the models, thus allowing our multi-layer Gaussian process prediction models to estimate not only their presence but also their severities. The experimental validation of our approach demonstrates strong agreement of the model predictions with these PD annotations. Our results show that the proposed method produces promising results in objective monitoring of movement abnormalities of PD in the presence of arbitrary and unknown voluntary motions, and makes an important step toward continuous monitoring of PD in the home environment.

A Human-Robot Interaction Perspective on Assistive and Rehabilitation Robotics.

Assistive and rehabilitation devices are a promising and challenging field of recent robotics research. Motivated by societal needs such as aging populations, such devices can support motor functionality and subject training. The design, control, sensing, and assessment of the devices become more sophisticated due to a human in the loop. This paper gives a human-robot interaction perspective on current issues and opportunities in the field. On the topic of control and machine learning, approaches that support but do not distract subjects are reviewed. Options to provide sensory user feedback that are currently missing from robotic devices are outlined. Parallels between device acceptance and affective computing are made. Furthermore, requirements for functional assessment protocols that relate to real-world tasks are discussed. In all topic areas, the design of human-oriented frameworks and methods is dominated by challenges related to the close interaction between the human and robotic device. This paper discusses the aforementioned aspects in order to open up new perspectives for future robotic solutions.

Robotic Billiards: Understanding Humans in Order to Counter Them.

Ongoing technological advances in the areas of computation, sensing, and mechatronics enable robotic-based systems to interact with humans in the real world. To succeed against a human in a competitive scenario, a robot must anticipate the human behavior and include it in its own planning framework. Then it can predict the next human move and counter it accordingly, thus not only achieving overall better performance but also systematically exploiting the opponent's weak spots. Pool is used as a representative scenario to derive a model-based planning and control framework where not only the physics of the environment but also a model of the opponent is considered. By representing the game of pool as a Markov decision process and incorporating a model of the human decision-making based on studies, an optimized policy is derived. This enables the robot to include the opponent's typical game style into its tactical considerations when planning a stroke. The results are validated in simulations and real-life experiments with an anthropomorphic robot playing pool against a human.

Rhythm patterns interaction--synchronization behavior for human-robot joint action.

Interactive behavior among humans is governed by the dynamics of movement synchronization in a variety of repetitive tasks. This requires the interaction partners to perform for example rhythmic limb swinging or even goal-directed arm movements. Inspired by that essential feature of human interaction, we present a novel concept and design methodology to synthesize goal-directed synchronization behavior for robotic agents in repetitive joint action tasks. The agents' tasks are described by closed movement trajectories and interpreted as limit cycles, for which instantaneous phase variables are derived based on oscillator theory. Events segmenting the trajectories into multiple primitives are introduced as anchoring points for enhanced synchronization modes. Utilizing both continuous phases and discrete events in a unifying view, we design a continuous dynamical process synchronizing the derived modes. Inverse to the derivation of phases, we also address the generation of goal-directed movements from the behavioral dynamics. The developed concept is implemented to an anthropomorphic robot. For evaluation of the concept an experiment is designed and conducted in which the robot performs a prototypical pick-and-place task jointly with human partners. The effectiveness of the designed behavior is successfully evidenced by objective measures of phase and event synchronization. Feedback gathered from the participants of our exploratory study suggests a subjectively pleasant sense of interaction created by the interactive behavior. The results highlight potential applications of the synchronization concept both in motor coordination among robotic agents and in enhanced social interaction between humanoid agents and humans.

Dyadic movement synchronization while performing incongruent trajectories requires mutual adaptation.

Unintentional movement synchronization is often emerging between interacting humans. In the present study, we investigate the extent to which the incongruence of movement trajectories has an influence on unintentional dyadic movement synchronization. During a target-directed tapping task, a participant repetitively moved between two targets in front of another participant who performed the same task in parallel but independently. When the movement path of one participant was changed by placing an obstacle between the targets, the degree of their unintentional movement synchronization was measured. Movement synchronization was observed despite of their substantially different movement trajectories. A deeper investigation of the participant's unintentional behavior shows, that although the actor who cleared the obstacle puts unintentional effort in establishing synchrony by increasing movement velocity-the other actor also unintentionally adjusted his/her behavior by increasing dwell times. Results are discussed in the light of joint action, movement interference and obstacle avoidance behavior.