Sammendrag
The purpose of this work is to review simple
classification schemes of the past and generalise these
to allow simultaneous, proportional control of multiple
prosthesis joints. In mathematical terms this involves the
definition of a continuous mapping f from a vector v(t) of
EMG parameters to a vector θ(t) of prosthesis states to
be controlled, e.g. torque, velocity or position setpoints.
In more common terms, we want to predict the states of
the (phantom) limb based on measured EMG signals, so
that the prosthesis can be made to behave accordingly.
The rationale is the hypothesis that the user will more
easily adapt to a simple and smooth control function,
thus achieving improved utilization of the prosthesis.
Vis fullstendig beskrivelse