Moreover, the computational time performance of the algorithm made clear that a real-time application is actually possible. The use of MLR models and subsequent Kalman filter for the prediction correction have produced noteworthy results. The successive analysis, implements a different approach of end effector trajectory reconstruction and deriving from it the kinematic that has to be applied to the robotic limb. In order to accomplish this result, this study is first undertaking a target-oriented classification employing SVM. Therefore, the aim of this thesis is to create a reliable system, for real-time applications, able to decode and then replicate, on a humanoid robot, the same motor task of a subject, using as an input only EMG signals. Instead, implementations of full robotics controls derived from biological signals, like EMG and EEG, are still not fully accomplished, despite the many attempts in the researching panorama. Humanoid robots can already autonomously perform complex tasks through human gesture and speech.
Even if the the idea of robots able to replicate human operations has already been proposed to the popular culture from decades, performing a coordinate human-like movements requires the use of advanced tools, like movement reconstruction and kinematics modelling. Currently, a wide range of applications in medicine take advantage of robots usage, but this relationship has not yet revealed its full potential.
0 kommentar(er)
