A fuzzy qualitative framework for robot intelligent connection

This paper proposes a novel framework in fuzzy qualitative terms for an attempt of attacking the robot intelligent connection problem. Robot intelligent connection is in the context of closing the gap between symbolic or qualitative functions and numerical sensing and control tasks through a generalized robot kinematics. First, fuzzy qualitative robot kinematics is revisited which provides theoretical preliminaries for the proposed robot motion representation. Secondly, a motion representation based on Gaussian mixture models is presented where Gaussian functions are combined to model a multimodal density of fuzzy qualitative kinematics parameters of a robotic end effector using clustering. Finally, simulation results in a PUMA 600 robot demonstrated that the proposed method effectively provides a two-way connection for robot representations used for numerical and symbolic robot tasks.