Efficient motion planning and control for robotic systems in dynamic situations

• Recheneffiziente Bewegungsplanung und -regelung für Robotern in dynamischen Situationen

Shahidi, Seyed Amirreza; Corves, Burkhard (Thesis advisor); Vidoni, Renato (Thesis advisor)

Aachen : RWTH Aachen University (2023)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2023

Abstract

The cooperation of several mobile and stationary robots in consensus is considered as one of the most important enablers for efficient, agile, and freely networked assembly systems. In this sense, the dynamic transformation of production lines requires especial modelling and control strategies for robotic manipulators, in order to provide them with reliable and feasible motions that are efficiently optimal. Therefore, the essential differences of the systems' models should be considered individually when planning their motions. At the same time, the modelling of the system should provide the control systems with an intuitive and unambiguous framework to guide them optimally through the planned motion.In the present work, optimal modelling strategies for multibody systems with tree-topology, of which open-chain robotic manipulators can be considered as a subcategory, are discussed and developed. The optimality of a modelling strategy is understood as computational efficiency, singularity-free representation, and compactness. To this end, a unit dual quaternionic representation of the system's configuration is presented and comprehensively discussed. Based on this representation, the geometry, kinetostatics, and dynamics of the multibody systems with tree-topology are developed. In order to facilitate the modelling and achieve an intuitive formulation, the geometric framework of screw theory is discussed on the analytical basis of Lie theory. This foundation can be employed to model control strategies that fulfil the premises of reliable and efficient motion control.In order to enable efficient motion planning for open-chain robotic manipulators, the essential differences between the cardinalities of their configuration manifold and their work space are considered, leading to the construction of a data structure in the form of a graph, dubbed Kinematic Graph. Using this graph, it is possible to efficiently apply the well-developed motion planning Algorithms, such as the sampling-based motion planning Algorithm, in the case study of robotic manipulators with open chains. These Algorithms were originally developed for low dimensional problems and are generally not suitable for the case study of robotic manipulators. Such Algorithms allow the application of planning strategies with theoretical optimality guarantees in the planning. This leads to the planning of an efficiently optimal motion for open-chain robotic manipulators with decoupled geometry, where the planning for the regional and the local structures should be conducted in conjunction with each other. Developments already exist for these optimal Algorithms to solve problems where conditions (such as the environment) are subject to change. The enabled application of these Algorithms to robotic manipulators makes it possible to plan motion for such systems in dynamic situations.

Institutions

• Chair and Institute of Mechanism Theory, Machine Dynamics and Robotics [411910]