Innovative interfaces for motor assessment

Motor abilities may be reduced in different conditions, such as neuromotor diseases, the physiological aging, or work-related musculoskeletal disorders. In the clinical realm, motor assessment is useful to measure the severity level, thus supporting physicians’ decision for diagnostic, prognostic, and rehabilitative purposes; on the other hand, an objective evaluation of the motor performance could allow for recording the exertion perceived by the subject while executing an industrial task. However, the clinical scales may suffer from subjectivity, since they are observation based and related to the specific background of different clinicians; the perceived exertion is conventionally estimated by self-ratings, which may be biased by the user’s psychology. Therefore, quantitative and objective measurement of motor abilities are needed to pursue more generalizable outcomes in both clinical and occupational applications. The purpose of this Ph.D. thesis is to illustrate the research works carried out during the conceptualization, design, implementation, and validation of frameworks for the quantitative assessment of motor capabilities by means of innovative interfaces based on serious game, deep-learning methods, and robotic exoskeletons. Serious games promote the engagement of the experimental subjects, thus keeping them motivated during the execution of multiple repetitions of the experimental tasks. Deep Learning models allow for the automatic recognition of motor patterns from raw data for a variety of applications, including human activity recognition and pathological gait recognition. Robotic exoskeletons can support humans in the execution of repetitive and exhausting motor tasks, thus preventing the injuries connected with work-related musculoskeletal disorders. The applications considered span from visuomotor adaptation to activity recognition and power augmentation. Tasks under consideration concerned the locomotion on a treadmill while controlling a virtual avatar, the execution of activities of daily living, as well as static and dynamic lifting tasks that are typical of an industrial scenario. Apaucity has been found in the different domains of the scientific literature to which the works presented in this thesis belong. As regards visuomotor adaptation, a few works implemented SGs to elicit sensorimotor learning in children during a walking task; therefore, more investigations are needed to perform a SG-aided evaluation of visuomotor adaptation capabilities of people in developmental age during locomotion tasks. With regards to human activity recognition, a minority of studies trained DL models with inertial data related to a separate execution of human motor actions and tested them with data acquired during an uninterrupted execution of the same activities; furthermore, there exist a few works exploiting simulated gait disorders to train DL models for recognizing pathological gaits. In the field of occupational exoskeletons, a gap has been found about the validation of such robotic devices with motor tasks resembling those of an industrial scenario with both conventional electromyographic measures and innovative methods based on graph theory. Therefore, the technical contributions of this thesis include the conceptualization of a locomotor task for the evaluation of visuomotor adaptation based on serious game; the validation of a framework based on deep-learning for the recognition of human activities executed in an uninterrupted sequence; the preliminary validation of a similar workflow addressing the recognition of mimicked gait disorders; the validation of an occupational exoskeleton assisting humans during industrial-like motor tasks by means of both traditional electromyographic measures and innovative approaches based on muscle networks. This thesis work is organized into two parts, each of which is divided in sections including an introduction and the works belonging to the specific context. More in detail, Chapter 1 is focused on applications for clinical purposes, giving an introduction of the objective and the technical contribution of the thesis in such context. Therefore, Section 1.2 describes the contributions proposed in the context of visuomotor adaptation assessment based on serious game, together with the related state-of-the-art. Sections 1.3 and 1.4 present the scientific literature and the contributions proposed in the context of activity recognition, concerning the classification of human motor actions performed continuously and pathological walking patterns simulated by healthy subjects, respectively. On the other hand, Chapter 2 is focused on applications for occupational purposes, giving an introduction of the objective and the technical contribution of the thesis in such context. Hence, Sections 2.2 and 2.3 report the state-of-art and the contributions proposed in the realm of the validation of occupational ex oskeleton with conventional electromyographic metrics and functional connectivity analysis based on muscle networks, respectively. Lastly, final remarks and considerations are drawn in Chapter 3