A muscle synergy-based method to improve robot-assisted movements.

There is increasing interest in using assistive robotic devices to support motor re-learning and recovery in individuals with neurological impairments. These robots aim to enhance overall motor control by providing adaptive assistance. However, using muscle synergies in designing control strategies for rehabilitation devices remains an emerging area of research. This study proposes a novel synergy-based objective function to assess how changes in robotic assistance levels affect muscle synergies in a 2D reaching task in the horizontal plane. Healthy participants performed the task in a transparent mode and with three levels of assistance while holding a weight. EMG signals from seven muscles were decomposed into muscle synergies across all conditions. First, we defined the three reference synergies as baseline knowledge of the synergies required to execute the task and their modulation, resulting in three main muscle recruitment strategies across ten participants. We then introduce three metrics to assess variations in motor coordination relative to the reference synergies. These metrics assessed how closely participants' muscle activation patterns matched the reference synergies, capturing variations in the shape, timing, and overall similarity of the muscle activation profiles. Finally, by combining the metrics, we present an objective function that assesses participants' motor coordination when performing the task with the added weight. The results highlight the importance of personalized assistance, as not all individuals could closely match the reference synergies with the same level of assistance. Additionally, the objective function demonstrated statistically significant differences in performance across assistance levels. Although this is a preliminary study, it presents promising results as a first step towards implementing human-in-the loop optimization in robotic-assisted rehabilitation.