||A new control system for elbow prostheses is proposed. The system would use myoelecuic
activity (MEA) meastuements to modulate the execution of a maxirnally smooth, minimum
jerk movement A prosthesis controller may extract information from MEA meastuements
about both the intended movement, for example elbow flexion, and about the time course
of the movement including, for example, average speed and duration.
Recent advances in myoelectric signal measurement have identified a deterministic
portion of MEA at the onset of movement This information has been used to determine
gross movement information, including movement direction and joint selection Common
prosthetic control schemes would then generate a movement with either constant velocity
or with the velocity proportional to measured MEA. These movements are simple in concept
but appear unnatural and are difficult to control during interactive tasks.
In contrast, a minimum jeik movement has interesting properties that make it a plausible
choice as a movement primitive Each minitnum jerk segment can be parameterized with
three par ameter s Commonly these are chosen to be start time, finish time and average
velocity. Researchers have proposed methods by which these movements can be combined
by straightfor ward super-position tecluilques to yield natural movements This paper
discusses minimum jerk movements as prosthesis movement primitives and presents a
control system architecture that uses deterministic information from MEA for movement
selection and MEA amplitude estimates for on-line modulation of movement parameters
Application of this type of control system should result in smooth prosthetic movements.