Advanced feeback system for myoelectrically controlled prostheses

The usefulness of prosthetic hands and arms for their wearers is presently limited by the size. and weight of the mechanical components and the power supply. The next limitation will be the control system, which lacks a data path from the prosthesis to the wearer which prevents it from reporting back to the wearer and can limit the flow of command data. This thesis describes a data path using electrocutaneous stimulation as the communication medium. The underlying premiss is that future control systems will be very sophisticated, with a large vocabulary of commands. To promote quick progress with the system, new wearers will have access to a few commands only. As expertise and demands develop, facilities can be unlocked and used without the expense, trouble and learning time of new prostheses The idea of using electrocutaneous stimulation is not new. Previous systems and some mechanical and other alternatives are reviewed. By a new approach to the quantisation and encoding of the data, the objections to the earlier systems are overcome. Parallel transmission of data was preferred for speed, but the poor spatial resolution of the nervous system prevents an electrode array small enough for a prosthetic environment. A new technique is described in which data in word or analogue fonn is presented as a trajectory through an array of sequentially energised electrodes. The wearer's needs impose time constraints for the presentation and recognition of words which are met. The controls required to ensure reliable signalling without discomfort or pain are investigated, and can be provided conveniently. Design criteria for the electrode array are given, based upon experimental observation. A suitable design is described. The trajectories need to be easily distinguishable for early success, so an analytical technique for assessing trajectories has been devised. The model is imperfect because the afferent nervous system is severely non-linear, and that non-linearity is not well understood, but one may write a set of well differentiated trajectories and add more as expertise is gained, with little risk of confusion. The information channel thus provided has several applications. These include feedback to improve control, the reporting of position, a substitute for tactile sensation, and facilitating the progressive development of expertise. The investigations are not yet relevant to lower limbs because electronic controls have only recently been found useful in them, and the command channel is not yet a design constraint