Information-based inverse kinematcs modeling for animation and robotics

The paper proposes a novel method for extremely fast inverse kinematics computation suitable for animation of anthropomorphic limbs, and fast moving lightweight manipulators. In the information intensive preprocessing phase, the workspace of the robot is decomposed into small cells, and joint angle vectors (configurations) and end-effector position/ orientation (posture) data sets are generated randomly in each cell using the forward kinematics. Due to the existence of multiple solutions for a desired posture, the generated configurations form clusters in the joint space which are classified. After the classification, the data belonging to each solution is used to determine the parameters of simple polynomial or neural network models that closely approximates the inverse kinematics within a cell. These parameters are stored in a lookup file. During the online phase, given the desired posture, the index of the appropriate cell is found, the model parameters are retrieved, and the joint angles are computed. The advantages of the proposed method over the existing approaches are discussed in the paper. In particular, the method is complete (provides all solutions), and is extremely fast. Statistical analyses for an industrial manipulator and an anthropomorphic arm are provided using both polynomial and neural network inverse kinematics models, which demonstrate the performance of the proposed method.