Myoelectric Control System and Task-Specific Characteristics Affect Voluntary Use of Simultaneous Control

Abstract

Clinically available myoelectric control does not enable simultaneous proportional control of prosthetic degrees of freedom. Multiple studies have proposed systems that provide simultaneous control, though few have investigated whether subjects voluntarily use simultaneous control or how they implement it. Additionally, few studies have explicitly evaluated the effect of providing proportional velocity control. The objective of this study was to evaluate factors influencing when and how subjects use simultaneous myoelectric control, including the ability to proportionally control the velocity and the required task precision. Five able-bodied subjects used simultaneous myoelectric control systems with and without proportional velocity control in a virtual Fitts' Law task. Though subjects used simultaneous control to a substantial degree when proportional velocity control was present, they used very little simultaneous control when using constant-velocity control. Furthermore, use of simultaneous control varied significantly with target distance and width, reflecting a strategy of using simultaneous control for gross cursor positioning and sequential control for fine corrective movements. These results provide insight into how users take advantage of simultaneous control and highlight the need for real-time evaluation of simultaneous control algorithms, as the potential benefit of providing simultaneous control may be affected by other characteristics of the myoelectric control system.

Fig. 1

Configuration for parallel dual-site control. An agonist/antagonist muscle pair was used to control the output velocity for each DOF. For each muscle, the MAV was calculated from the EMG and was conditioned by a linear gain and threshold. When proportional velocity control was present, the difference in the conditioned signals determined the velocity of the DOF (V_Prop) by Eq 1. When proportional velocity control was not present, V_Prop was transformed to a constant velocity output (V_Const) by Eq 2. The muscles used included a pair of rotators: pronator teres and supinator; a wrist flexor and extensor: flexor carpi radialis and extensor carpi radialis longus; and a finger flexor and extensor: flexor digitorum profundus and extensor digitorum communis.

Fig. 2

Fitts’ Law analysis of user performance with both myoelectric control systems. Empirical Fitts’ Law regression models for (A) proportional velocity control and (B) on-off control. A linear relationship was observed between completion time and index of difficulty for all three target complexities. (C) Difference in throughput between the two control systems. Percent differences reflect the increase in throughput observed when using proportional velocity control compared to on-off control, and were averaged across subjects. Significantly greater throughput (p < 0.05) was present when subjects had proportional velocity control for all target complexities. (D) Difference in path efficiency between the two control systems. Percent difference reflects the increase in throughput observed when using proportional velocity control compared to on-off control. Significantly greater path efficiency (p < 0.05) was present when subjects had proportional velocity control for all target complexities. Error bars represent standard error (N=5).

Fig. 3

Average use of simultaneous control versus normalized trial duration for (A) proportional velocity control and (B) on-off control. Subjects used significantly more simultaneous control when proportional velocity control was also available (p<0.001). When proportional velocity control was available (A), subjects primarily used sequential control of a single DOF for 1-DOF targets. For 2- and 3-DOF targets, subjects’ use of simultaneous control peaked in the first half of the trial duration. Subjects used more sequential control during the second half of the trials. Shaded region represents standard error (N=5).

Fig. 4

Average use of simultaneous control as a function of target complexity and distance and width parameters. Subjects used more simultaneous control for targets that were farther away from the starting position and for targets with greater annulus widths. This general trend held across all three target complexities. Error bars reflect standard error (N=5).

Fig. 5

Cursor dynamics as subjects used simultaneous control versus sequential control. (A) For all three target complexities, on average, subjects used simultaneous control when they were farther from the target (p < 0.001). (B) For all three target complexities, subjects moved the cursor at greater velocities when they used simultaneous control than when they used sequential control (p<0.001). Error bars represent standard error (N=5).