| This dissertation addresses manipulator redundancy from a global perspective, aiming at the kinematic control through the exploration of self-motion topology. Our methodology is based on collecting information about the structure of the kinematic map--like a road map--with the use of topological tools, providing an overall view of the configuration space and its relationship to the work space--a suitable framework for the efficient implementation of global approaches.;The introduction of kinematic constraints has a significant impact onto the self-motion topology. The analysis of the effect of those constraints led to the discovery of a new type of singularity--the semi-singularity--which is unidirectional in nature, as opposed to the traditional singularity. The semi-singularity is the key factor for the reshaping of the global kinematic map due to kinematic constraints.;A space discretization method has been developed to benefit from the topological structure, embedding kinematics in its representation. This method enables an efficient exploration of global redundancy resolution and path planning, thus avoiding local minima and deadlocks with minimum effort.;The structure derived from the topology analysis is suitable for the implementation of a top-down approach for the kinematic control, exploring hierarchical planning and parallelism, in an effort to overcome the added complexity of a redundant system. Furthermore, it offers a framework for human intervention and interaction with the kinematic control at the highest abstract level through graphics displays.;The discretization and hierarchical planning were implemented for a planar manipulator, demonstrating significant improvement in the search for globally optimum solutions of path planning when compared to traditional approaches such as the configuration space regular grid.;As a case study, the 8-DoF AAI Arm was analyzed, demonstrating the feasibility of the method for industry-scale manipulators.;While the core contribution of this dissertation is conceptual in nature, it provides examples, implementations and a case study to illustrate the applicability of our methodology and the potential for great future advances in the kinematic control of redundant manipulators. |
