Abstract : After the transmission of image and sound through networks, it will be relevant to be interested by the transmission of the gesture. The “3D touch” is making a revolution in the way of working for the 3D users on computers. It allows people to directly manipulate objects or numerical data in the same way as in reality by using their feel of touch. Some force feedback devices seem like handles with small engines driven by software drivers. They return back to the manipulator's hand feelings of forces or vibrations, they are computable in function of task you want to make in the virtual word. During a distant touching, a telegesture, we are faced to the problem of the instability of the haptic loop (human-force feedback device- force feedback in the hand of the operator). This instability is due to the delay introduced by the uncompressible latency of network. Ideally we would have an instantaneous response (inferior to one millisecond), but the working out of such response is unfeasible cause to the non- causality of this. In order to deal with the delay introduced by the network, we have used predictors to be able to obtain a simultaneous response to the excitation. This ergonomic improvement of telegesture is realized thanks to the statistical relevance of the model of the predictive algorithm.