Observer approach for fault detection of the handwriting process

This paper deals with a fault detection approach based on an observer, applied to the handwriting biological process. This process is characterized by a linear model generating handwritten ships on bi-dimensional plane from two muscles activities of the forearm, called ElectroMyoGraphic signals (EMG). The specificity of our work is that the inputs of the studied system are unpredictable. This implicates complexities for the diagnosis of different faults that can be encountered during the handwriting motion. The actuator of this biological process is sensitive to several faults from different origins (motor disability of the hand, neurological disease, problem in the sensors of EMG signals or in the transmission cable, etc.). These faults can produce undesirable fluctuations in the graphic trace generated by the support system made to reconstruct handwritten shapes as produced by a human. In this sense, we are interested on studying these fault types and the applying of an observer-based fault detection approach to figure out a fault occurrence.