Fig. 5: Shared control improved grip security and reduced cognitive burden on a holding task.

A Six intact-limb participants with no prior myoelectric control experience completed a holding task³⁰. Participants attempted to pick up a spherical ball with a power circular grasp and hold it for two minutes without dropping it. Participants had independent control of the thumb, index, middle, and ring fingers. All other parameters were the same. B Example data from the thumb during the holding task with shared control. The participant began by extending their fingers (dashed line) to wrap around the object (~0.5 to 1 s), and then flexed their fingers to grasp the object (~1 to 1.75 s). Once the object was within the grasp, the machine control (dotted line) provided a steady offset to the shared output (solid line) so the fingers made contact with the object (~1.75 s onward). Shared control effectively maintained a stable grip even when the human contribution varied and decreased over time (~2.5 to 6 s). With shared control, participants dropped the ball significantly less (permutation test; p = 0.037; C) and were able to hold the ball continuously for significantly longer (permutation test; p = 0.039; D). Participants tended to have more fingers making contact with the object (E), although this effect was not statistically significant. Shared control resulted in significantly less cognitive burden (generalized linear model; p = 0.016; F). Participants tended to exert less physical effort when using shared control, although this effect was not statistically significant (G). Violin plots represent the distribution of the data; black bars indicate the median. Data from six intact-limb participants. Sample sizes are: 12 drop counts per condition (2 repeated measures per participant); 12 maximum hold times per condition (2 repeated measures per participant); 12 average number of finger measurements per condition (2 repeated measures per participant); 398 DRT response times for human control and 468 DRT response times for shared control; 12 sEMG MAV averages per condition (2 repeated measures per participant). * denotes p < 0.05. Numerical results are found in Supplementary Table 1. All statistical tests are two-sided.