A Fuzzy Qualitative Framework for Connecting Robot Qualitative and Quantitative Representations

This paper proposes a novel framework for describing articulated robot kinematics motion with the goal of providing a unified representation by combining symbolic or qualitative functions and numerical sensing and control tasks in the context of intelligent robotics. First, fuzzy qualitative robot kinematics that provides theoretical preliminaries for the proposed robot motion representation is revisited. Second, a fuzzy qualitative framework based on clustering techniques is presented to connect numerical and symbolic robot representations. Built on the
k
-\bb
AGOP
operator (an extension of the ordered weighted aggregation operators), k-means and Gaussian functions are adapted to model a multimodal density of fuzzy qualitative kinematics parameters of a robot in both Cartesian and joint spaces; on the other hand, a mixture regressor and interpolation method are employed to convert Gaussian symbols into numerical values. Finally, simulation results in a PUMA 560 robot demonstrated that the proposed method effectively provides a two-way connection for robot representations used for both numerical and symbolic robotic tasks.