Robots as cyber-physical systems: new perspectives for the design of autonomous robots

Today’s control systems are made of an embedded system connected to the system to be controlled through a feedback loop; thus the physical process influence the computation and vice versa. The term “cyber-physical system” defines a control system developed according to a new design methodology, which models both the systems taking into account their reciprocal interactions. In the current scientific literature, the design of control systems for au- tonomous robots is a problem to be addressed separately from the application of real-time techniques or artificial intelligence techniques on them. However, choices made to solve one of the two problems may have consequences on the other. Therefore, it seems that it is not sufficient to separately understand both the problems: it is important to understand the interaction between them. This thesis proposes an alternative approach to the design of intelligent au- tonomous robots, derived from the cyber-physical approach to control systems, which consists in tackling jointly both the problem of controlling the robotic system and the problem of deploying an artificial intelligent application on it. In the first part of the thesis various modelling abstractions are presented and described in detail. Continuous and discrete modelling techniques for physical and computing systems are described in chapters 2 and 3, respectively. Chapter 4 describes modelling techniques to integrate the previous models in a single model. Chapter 5, then, introduces some modelling techniques for concurrent computation. The second part of the thesis is devoted to embedded and real-time systems. In chapter 6 the most common embedded architectures are described, focusing on parallelism and on the characteristics that make them suitable or unsuitable for application in intelligent autonomous robotics. Chapter 7 introduces vari- ous I/O mechanism and the interrupt paradigm, which governs the interaction between high-level computing systems and low-level control systems. Chap- ter 8 describes the basic of real-time operating systems, focusing mainly on scheduling under various degree of constraints. Finally, the third part of the thesis apply theories and methodologies intro- duced in the first two part to the design of an intelligent autonomous robot system to be used in the contexts of video-surveillance and ambient-assisted living.