Fig 1.
Experimental setup and vibrator positioning.
Panel A: Participants sat medially to the robot and held a custom plastic handle with their right hand. Another custom plastic support was attached to the robot EE. Passive markers were attached to both handle and support to track them using infrared cameras. Participants moved the handle upon the transparent table while the robot EE moved below the table. The handle was equipped with a soft pad to reduce friction with table surface. Participants wore disposable earplugs to mask the noise of the robot moving, and could be blindfolded to remove visual feedback, depending on the experimental block. Panel B: A total of four vibrators were worn by participants: two of them on the medial (2A) and lateral aspect (2B) of the thigh, about 12 cm proximal to the knee; the other two on the frontal (1A) and posterior aspect (1B) of the calf about 20 cm proximal to the ankle. Vibrators were placed to be in touch with soft tissue and avoid bone crest. The matrix showing information conveyed by vibrators in each condition is shown on the right. X and y represent the EE coordinates with respect to the workspace centre, while J1 and J4 represent the first and fourth robot joint angles, as represented in the lower part of panel B. Positive and negative values of x, y, J1 and J4 are considered with respect to the starting configuration of the robot, with the EE located in the centre of the workspace (x, y, J1 and J4 equal to 0). The lower part of panel B shows which vibrators were activated when joints rotated (Joint-space feedback, green) or when the EE moved away from the workspace centre (Task-space feedback, yellow).
Fig 2.
Representation of the half-circle shaped workspace.
Panel A. P1, P2 and P3 represent three sequential, randomly generated, target points. C represents the centre point, which is both the starting and the ending position of a closed path. Lines 1, 2, 3 and 4 represent straight motions of the robot from one point to another. Panel B. Aiming direction (light blue) of the participant computed as the vector linking the starting point and the average 25% of the whole human path (dotted grey), robot path (red) and aiming error (green) computed as the angle between human and robot direction vectors.
Fig 3.
“Visual” and “Vibrotactile” refer to the type of feedback relayed to participants (blue shade), while “Participant’s Movement” refers to participants’ arm movement which followed robot motions. Numbers below experimental blocks indicate the number of trials and express how many times the closed paths of 4 sequential motions were repeated within a specific block. Green ticks denote presence of a feedback type/movement, while red marks denote their absence. Green semi-transparent tick in Training C denotes presence of Visual feedback only at the end of each motion, to allow the correction of the position.
Fig 4.
Position error, onset delay and aiming error.
Position error obtained by participants during baseline (A) and test block (B), onset delays obtained in baseline (C), delay compensated position error obtained during test (D) and aiming error obtained during baseline (E) and test block (F). Joint-space feedback condition is shown in green while Task-space feedback condition is shown in orange. Red line represents median values. * denotes p<0.05, ** denotes p < 0.01.
Fig 5.
Each bar represents the position error obtained by all participants during a single Closed path, computed as the mean of the position error of four sequential motions. Red lines represent median value and bars are filled within 25th and 75th interquartile range. Panel A shows learning index during Joint Space (green) condition while Panel B refers to Task Space (orange) condition. Blue lines represent the linear regression.