Navegar por los elementos (1 total)
- Resumen es exacto "Cyber-Physical Systems (CPS) integrate the physical and algorithmic domains for which compliance of the quality of operation must be controlled both on physical and cybernetic processes that influence each other. They are hybrid systems, whose main dynamics require formal representations as continuous, discrete time and/or discrete event systems. For CPSs, a comprehensive quality of operation must be guaranteed even in unpredictable operating environments. For instance, applications such as Unmanned Aerial Vehicles (UAVs) can suffer catastrophic consequences if the quality of control of the physical dynamics is inefficient. If we also consider that the mission of a UAV involves data processing during the flight, a poor control of the quality of service of the software on-board can lead to the failure of a whole mission. The context tends to be one of scarce computational resources, with robotic platforms being a paradigmatic example, where the dynamic and efficient allocation of resources to different controllers becomes a central aspect. Therefore, it becomes desirable to provide the CPS with self-adaptive capabilities through cooperative controllers. For example, flight stability can be temporarily relaxed in favor of the efficiency of a communication protocol, and vice versa. This is the field of action of hybrid controllers, which seek to satisfy multiple control objectives in systems modeled by continuous dynamics (e.g. differential equations) and discrete dynamics (e.g. state machines). Currently there is no clear and unified theory that allows the design of hybrid controllers under the same analysis framework, allowing to capture the interaction between physical and computational requirements. This calls for heterogeneous controller design processes under a unified platform. Modeling and simulation engineering can provide the necessary framework to facilitate the robust interconnection of hybrid components. In this Thesis we develop new tools for modeling, simulation and validation of hybrid controllers for CPS with the ability to satisfy dynamic resource allocation requirements. We adopt the DEVS (Discrete Event System Specification) modeling and simulation formalism that allows us to represent exactly any discrete system and to approximate continuous systems with any desired accuracy level, thus obtaining hybrid models with guarantees of simulation correctness. Then we propose a design methodology based on model continuity. This allows to develop control algorithms in the form of DEVS simulation models that are preserved and refined during the development cycle including their final deployment in robots, where they operate in hardware-in-the-loop mode. We also introduce new supervisory, hybrid and hierarchical controllers, which offer comprehensive service quality guarantees in UAV-type robots where flight controllers and software systems compete for scarce resources. The case studies include field experimentation with multirotor vehicles, fixed-wing aircraft, and an hovercraft robot."
Ordenar por: