Personalized Robotic Training on a Planar Reaching Task with Simulated Stroke

Based on recent advancements in neuro-adaptive control, we conduct an experiment to eval- uate a novel iterative algorithm for generating customized training forces. The objective is to perturb the subjects movements during the reaching task execution with two different types of robot-generated forces, resulting in an evidence of motor adaptation induced to compen- sate these perturbations. The main hypothesis is that the adaptation process would induce changes in the feed-forward command that would result in movements’ trajectories alteration. The results indicate that after a training period with robot-generated forces, the hand trajecto- ries undergo modifications due to internal model adaptation, leading to improved performance measured in terms of position error. The experiment explores motor learning in two sets of directions and compares two conditions: one with only the curl force field, which is the first type of force, and the other adding a custom Error Field force. The findings highlight the effectiveness of the Error Field force, as the group subjected to this force demonstrates enhanced motor learning, characterized by a higher and faster decrease in error across trials. These results showcase the potential of these techniques to induce motor improvements for personal activities or neurorehabilitation purposes.