We introduce a kinematic graph in this article. A kinematic graph results from structuring the data obtained from the sampling method for sampling-based motion planning algorithms in robotics with the motivation to adapt the method to the positioning problem of robotic manipulators. The term kinematic graph emphasises the fact that any path computed by sampling-based motion planning algorithms using a kinematic graph is guaranteed to correspond to a feasible motion for the positioning of the robotic manipulator. We propose methods to combine the information from the configuration and task spaces of the robotic manipulators to cluster the samples. The kinematic graph is the result of this systematic clustering and a tremendous reduction in the size of the problem. Hence, using a kinematic graph, it is possible to effectively employ sampling-based motion planning algorithms for robotic manipulators, where the problem is defined in higher dimensions than those for which these algorithms were developed. Other barriers that hindered adequate utilisation of such algorithms for robotic manipulators with articulated arms, such as the non-injective surjection of the forward kinematic function, are also addressed in the structure of the kinematic graph.
The increasing product variability and lack of skilled workers demand for autonomous, flexible production. Since assembly is considered a main cost driver and accounts for a major part of production time, research focuses on new technologies in assembly. The paradigm of Line-less Mobile Assembly Systems (LMAS) provides a solution for the future of assembly by mobilizing all resources. Thus, dynamic product routes through spatiotemporally configured assembly stations on a shop floor free of fixed obstacles are enabled. In this chapter, we present research focal points on different levels of LMAS, starting with the macroscopic level of formation planning, followed by the mesoscopic level of mobile robot control and multipurpose input devices and the microscopic level of services, such as interpreting autonomous decisions and in-network computing. We provide cross-level data and knowledge transfer through a novel ontology-based knowledge management. Overall, our work contributes to future safe and predictable human-robot collaboration in dynamic LMAS stations based on accurate online formation and motion planning of mobile robots, novel human-machine interfaces and networking technologies, as well as trustworthy AI-based decisions.