A Neural Network Based Sliding Mode Control for Free-Flying Space Manipulators

The capability of regulating the attitude and position of spacecraft equipped with a robotic manipulator is crucial to achieve the practically un-limited work space. This paper presents an innovative neural network based less chattering sliding mode control (NNLSMC) scheme in which a Radius Basic Function (RBF) neural network is used to approximate the lump uncertainties. The proposed new adaptive law for the switching gain is positively proportional to the absolute value of sliding variables during the initial period, while the adaption law is then switched to be the designed tangent function of absolute value of sliding variables after entering into the vicinity of sliding manifold. To ensure the stability, the RBF neural network with bounded parameters is used. The tracking errors of both the manipulator's joints and spacecraft's postures (attitude and position) are proved to be uniformly ultimately bounded under the proposed control scheme, and less chattering effects on both the control inputs and sliding variables are achieved.