Software framework of control systems on an MPSoCs platform

• Ano de defesa: 2023
• Autor(a) principal: Vancin, Paulo Henrique
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul; Escola Politécnica; Brasil; PUCRS; Programa de Pós-Graduação em Ciência da Computação
• Palavras-chave em Português: Control Systems; Embedded Systems; Heterogeneous Computing; Robotics; MPSoC; Sistemas de Controle; Sistemas Embarcados; Computação Heterogênea; Robótica; CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO
• Link de acesso: https://tede2.pucrs.br/tede2/handle/tede/10792

With the increasing complexity of robotic systems, many aspects of their control system architecture also become more complex. Sensing produces huge data aggregates to collect and process; actuators demand rapid signal manipulation, and controllers evolve to include highly complex algorithms. This progression in processing demand requires computing power to keep up. However, new processor technologies introduce power limits. These limits, implies that the dissipation of energy inside the chip prevents all its resources from being used simultaneously at their maximum performance rate. With this physical limitation, a distinct method is needed to continue to increase hardware performance. One way to deal with such high processing demands is through the use of heterogeneous computing. A heterogeneous computing system distributes data, processing, and program execution across different processors. The basic idea of this thesis is that the application of a digital control system in a heterogeneous computing system increases the efficiency of the controller, while allowing multiple techniques to be added to the control. Examples are the decentralization of the control architecture, the self-adaptation of the controller, fault tolerance techniques and the energy management. This thesis actually proposes a software framework for the implementation of control systems in a multiprocessor embedded system (MPSoC). This framework was developed in a generic way to serve multiple robotic artifacts. The application case study employed herein is a quadrotor unmanned aerial vehicle. Such an example of robotic equipment is considered due to its fast dynamics, its sensitivity to faults and its high demand for energy management and powerful controllers. The software framework was embedded in a simulation environment capable of simulating both the processor and the quadrotor. Six sets of experiments validate the general hypothesis of the thesis. These experiments tested the time requirements, the decentralization of the control architecture, the ability to process complex control algorithms without impacting the performance, the intra-chip fault tolerance, not forgetting the power management applications and the self-adaptation of controllers