A Framework For Kinematic Modeling of Cooperative Robotic Systems Based on Screw Theory

This paper concerns the analysis of the screw theory-based kinematic modeling in order to ease the programming process. To do so, an object-oriented computational framework is developed from this analysis. Screw theory and related tools are well used in motion analysis of open and closed kinematic chains, typical of robotic manipulators, in an uniform and systematic way. Their use in the inverse kinematics resolution of redundant systems present advantages such as the dimensional consistency and the tendency to conserve movement. These are of interest in motion planning strategies for systems where redundancy is an inherent and necessary feature, such as parallel manipulators, cooperative robotic systems and vehicle-manipulator systems. These systems can present complex kinematic chains. However, these chains can be decomposed into simpler ones, which can be previously derived and stored in a sort of kinematic chain database. Thus, complex kinematic chains can be formed by composition of predefined kinematic chains connected to each other, considering also the necessary transformations to express all their screws in a common reference system. Besides facilitating the kinematic model definition of complex chains, this systematization is interesting to deal with systems whose kinematic model can vary over time, as in collision avoidance situations. Therefore, the study of the kinematic modeling modularization is of interest to make kinematic analysis easier and even to automate this process. It is also noted that, to the best knowledge of the authors, the mathematical software usually used in kinematic analysis do not have modules/libraries to deal with screw theory entities. To implement the computational system modularity features, object-oriented analysis techniques were used to design a framework to represent screw theory entities, kinematic chains and their composition. This framework is intended to make the implementation of kinematic chain databases and the future development of an automated system friendly. The first results of this analysis are presented in this paper. The modularization of kinematic chains, the possibility of chain databases and the object-oriented model resulting from the analysis of robotic systems are discussed. Some criteria and restrictions to the choice of the computational platform used and an example are presented.