UDC 519.246 © B.G. Kukharenko, D.I. Ponomarev BAYESIAN FILTERING OF CONTROL SIGNAL OF TELEROBOTIC MANIPULATOR WITH PRICISE ACCELEROMETER A telerobotic manipulator is under study, which transmits to object under control an operator hand motion in three dimension (3D) space. <...> Under the manipulator development, a problem of filtering control signal arises, since a noise is produced by operator hand tremor process. <...> So the Kalman smoother is applied with linear system parameter estimates by the Expectation-Maximization algorithm. <...> Keywords: telerobotic manipulator, precise accelerometer, control signal, Bayesian filtering, Kalman smoother, Expectation-Maximization algorithm. 1. <...> However, telerobotic manipulators are under development recently, which transmit to the object under control an operator hand motion in tree dimension (3D) space. <...> The telerobotic manipulators are robotic systems in surgery, high sensitive joysticks, computer mouse etc all without a supporting surface and pointers. <...> Under the manipulator development, a problem of filtering control signal arises, since a noise is produced by the operator hand tremor. <...> The noise influence on control signal can be significant. <...> In present paper the problem is studied in case of a telerobotic manipulator, sensor of which is the precise three axes MEMS (Micro-ElectroMechanical System) accelerometer. <...> The acceleration projection analog values are selected periodically and converted to digital sequences ax(n), ay(n), az(n). <...> Next microcontroller filters the digital sequences and converts these to signals of positioning computer mouse. 2. <...> Spectral characteristics of control signal and noise Acceleration records of two sensitive axes of the accelerometer under study are obtained with timediscretization step equals 5 ms (Fig. 3). a Fig. 1. <...> Three sensitive axes (x,y,z) of MEMS accelerometer and acceleration of gravity g As an accelerometer have a sensitivity to force of gravity, varying manipulator placement with respect to acceleration of gravity g produces a variation of the acceleration projections on the accelerometer sensitive axes x, y, z (Fig. 1). <...> As example, the projections can be in use to positioning a cursor of computer mouse. <...> Acceleration records: a - shows gravity acceleration projection on axis x of accelerometer; b - shows gravity acceleration projection on axis y of accelerometer To estimate an importance of filtering noise n Fig <...>