A framework for modular and distributable control of reconfigurable robotic systems

In this paper, a distributable control approach for reconfigurable robotic systems is presented. The approach is based on the dynamic programming technique to solve the inverse kinematics problem in a distributed way. This algorithm is extended by an additional control input to reduce the orthogonal error to the desired Cartesian trajectory and to ensure smooth joint angles and velocities. The required torques to track the desired Cartesian trajectory are then computed by application of the well-known recursive Newton-Euler algorithm. Furthermore, the entire algorithm is implemented according to the international standard IEC 61499 and encapsulated into several reusable function blocks for controlling a single joint. This reduces the effort for implementing the control algorithm as well as the computational time, since they can be transferred to the corresponding joint control device. The evaluation of the algorithm is carried out on a simulation of a manipulator with four degrees of freedom.