Passive Magnetic Localization Based On Connotative Pre-Calibration For Tongue-Machine-Interface

In human-machine-interface study, recognition of conscious human motion by contactless passive magnetic marked method can provide abundant information, such as PM marked tongue-machine-interface (TMI). However, solution of this nonlinear magnetic inverse problem heavily relies on initialization. This paper takes advantage of the motion characteristics constrained by physiological structure, and develops an enhanced algorithm for real-time full-pose passive magnetic localization in TMI application. In the proposed algorithm, a predetermined-discretized database is introduced, and provides reliable initializations for the nonlinear magnetic localization problem. This database presents the connotative relationships between the location and orientation of passive magnetic source, termed location-orientation mapping database (LOMD). The influence of initialization and LOMD constitution on solving the nonlinear inverse magnetic problem are studied through extensive simulations. Test bench evaluations are conducted on the designed experiment system. The experiment results show that about 38.5 % localization improvement can be achieved with a real-time tracking frequency up to 56.5 Hz.