The research presented in this thesis investigates the viability of utilising a mobile robot as a substitute for faulty static sensors exhibiting anomalous behaviour at a door in a Smart Environment. This is implemented through the development of finite state automata for the management of static sensor events. Novel approaches for the introduction of adaptability to finite state automata and the investigation of static sensor anomalies are proposed. This research was executed through the use of both practical and theoretical investigations. The ideas proposed in this work were developed as software artefacts. The functionality of that software was tested in order to assess the validity of the approaches proposed within this body of work. The results of this research show that the utilisation of a mobile robot in a Smart Environment is a viable means of providing substitution for static sensor functionality about a door and can be used as a means of verifying the presence of a systemic static sensor anomaly. Adaptive finite state automata were successfully developed and validated as a means of implementing the self-healing property of the Autonomic Computing paradigm in a software system. The limitations of this research have been identified and considered in the conclusion of this thesis as areas for future work.

Being the cornerstone of today's Industry, Industrial Control Systems (ICSs) play an important role in the overall function and quality of modern society. Their use to control critical processes within the Industry (power production, transportation, manufacturing etc.) makes them an integral part of the Critical National Infrastructure (CNI), as defined by the European Council (2008), rendering thereby their protection a process of critical importance. Traditionally, ICSs have been operated as closed, isolated systems. However, the connection of contemporary ICS installations with external networks, including the corporate network and the Internet, along with the introduction of conventional off-the-self technologies, has exposed the once isolated systems to a rapidly evolving yet new to them cyberthreat landscape. Their critical nature further complicates the situation making them an attractive target for various attack vectors and threat agents. Traditional cyber-security methods seem inadequate since they are tailored to the specific corporate needs ignoring the demanding nature of ICSs. Nevertheless, due to the increased cost of designing and applying new cyber-protection methods, the majority of cyber-security solutions used nowadays in the Industry are mainly adaptations of traditional corporate-oriented methods (Giannopoulos et aI., 2012), raising thus significant challenges that the research community has to address. This thesis presents novel cyber-security approaches, tailored to the particular nature of ICSs. The developed methods take into account both the increased cybersecurity needs in this critical area and the related costs, offering optimal cost efficient cyber-security solutions. Stafford Beer's Viable System Model (VSM) was used as a vehicle to analyse the behaviour of ICSs and identify the areas where cost-efficient cyber-protection methods are in need (Stafford, 1984). Driving the research into those areas a series of cyber-protection methods were developed using system theory-base.d techniques such as game theory and system dynamics. Those methods include a cost-efficient cyber-security model against the malware spread within ICSs and a cost-efficient model for the protection against Denial of Service (DoS)/ Distributed Denial of Service (DDoS) attacks. Building on the same premises a novel ICS-oriented cyber-security risk management method was developed based on the Viable System Model and game theory. i

This thesis is concerned with the challenge of creating a robot which is capable of natural, verbal communication with humans. More specifically, it considers the task of categorising objects according to sensory stimuli. We focus on tactile texture perception, a sensory feature which has received relatively little attention in artificial intelligence, in comparison with vision and audition. Through a multidisciplinary approach involving sensory feature extraction, computer simulations, and psychophysical experiments, we compare texture perception and categorisation between human and robot, and consider the problem of enabling communication between them. We begin by presenting TacTip, an artificial fingertip sensor which we apply to the task of texture recognition. We describe a feature extraction process used to specify a textural feature space for the sensor, which is then used for texture recognition and categorisation. Next, we present a framework for robotic communication and learning. This framework consists of two main parts, the first of which is the representational model used by the robot to categorise perceived stimuli. We present a model based on random set theory and prototype theory, and compare this with a similar model based on Bayesian statistics. The second part of the framework is the context in which the robots communicate and learn. In our case this consists of a multi-agent simulation in which robots communicate with each other through pairwise interactions called language games, and thereby develop a shared set of categories. Finally, we consider how our robot might communicate with and learn from humans. We describe two psychophysical experiments, the first of which studies how humans naturally classify textures, the second investigating whether humans can learn specific categorisations presented to them. Each experiment can be interpreted as one part of a language game interaction between human and robot. We discuss our results in the context of human-robot communication.

Humans are extremely adept at interacting with the world through touch and eliciting meaningful information through touch. Current handheld and wearable electronic devices provide very limited tactile feedback to the user. The ultimate tactile device should be able to stimulate the full range of mechanoreceptors in the skin, to elicit any sensation. Our skin is soft and compliant, so wearable devices should be soft, conform to and move with the user. Smart materials could hold the key to development of low-cost high-resolution soft tactile devices. This thesis is working towards a vision of entirely soft and compliant tactile devices using smart materials. Laterotactile stimulation, where tactile elements move parallel to the skin surface to create regions of stretching and compression in the skin, would seem perfectly suited for such applications. Although there are examples of laterotactile stimulation in the literature, this area is largely unexplored. We present a psychophysical study comparing the tactile sensitivity to normal and lateral stimuli. We present and characterise a compliant laterotactile display prototype for stimulation of the glabrous skin, using Dielectric Elastomer Actuators. This technology will readily scale up to large arrays and large devices and scale down to micro-stimulators. Real-world tactile interaction often conveys inherent meaning and affect. Taking inspiration from recent work in neuroscience, we have developed the Tickler; a soft laterotactile display that strokes the wrist. The Tickler creates natural-feeling sensations, and highlights opportunities for soft technologies in haptics. We show that the perceived sensation of affective haptic devices can be modulated by the context in which it is presented, and the nature of the modulation is stimulus dependent. This has significant implications for the evaluation of affective haptic devices. We consider how follicular structures in Nature can be mimicked in robotic skin and haptic devices, and present a first example of a follicular skin device for two-way gesture-based tactile communication. These developments provide the important foundations for new devices that mark a paradigm shift in our physical interactions with technology and the virtual world.

No description available.

No description available.

This thesis proposes an outline for a framework for an evaluation method that takes as an input a model of an artificial hand, which claims to be anthropomorphic, and produces as output the set of tasks that the hand can perform. The framework is based on studying the literature on the anatomy and functionalities of the human hand and methods of implementing these functionalities in artificial systems. The thesis also presents a partial implementation of the framework which focuses on tasks of gesturing and grasping using anthropomorphic postures. This thesis focuses on the evaluation of the intrinsic hardware of robot hands from technical and functional perspectives, including kinematics of the mechanical structure, geometry of the contact surface, and functional force conditions for successful grasps. This thesis does not consider topics related to control or elements of aesthetics of the design of robot hands. The thesis reviews the literature on the anatomy, motion and sensory capabilities, and functionalities of the human hand to define a reference to evaluate artificial hands. It distinguishes between the hands construction and functionalities and presents a discussion of anthropomorphism that reflects this distinction. It reviews key theory related to artificial hands and notable solutions and existing methods of evaluating artificial hands. The thesis outlines the evaluation framework by defining the action manifold of the anthropomorphic hand, defined as the set of all tasks that a hypothetical ideal anthropomorphic hand should be able to do, and analysing the manifold tasks to determine the hand capabilities involved in the tasks and how to simulate them. A syntax is defined to describe hand tasks and anthropomorphic postures. The action manifold is defined to be used as a functional reference to evaluate artificial hands’ performance. A method to evaluate anthropomorphic postures using Fuzzy logic and a method to evaluate anthropomorphic grasping abilities are proposed and applied on models of the human hand and the InMoov robot hand. The results show the methods’ ability to detect successful postures and grasps. Future work towards a full implementation of the framework is suggested.

An investigation has been made into hybrid systems which include stochastic and deterministic optimization. This thesis aims to provide new and relevant insights into the design of the nature-inspired hybrid optimization paradigms. It combines evolutionary and gradient-based methods. These hybrid evolutionary methods yield improved performance when applied to complex global optimization tasks and recent research has shown many of such hybridization policies. The thesis has three broad contributions. Firstly, by examination of stochastic optimization, supported by case studies, we utilised the Price's theorem to formulate a new population evolvability measure which assesses the dynamical characteristics of evolutionary operators. This leads to the development of a new convergence assessment method. A novel diversity control mechanism that uses heuristic initialisation and convergence detection mechanism is then proposed. Empirical support is provided to explicitly analyse the benefits of effective diversity control for continuous optimization. Secondly, this study utilised research relevance trees to evolve hybrid systems which combine various evolutionary computation (EC) models with the sequential quadratic programming (SQP) algorithm in a collaborative manner. We reviewed the convergence characteristics of various numerical optimization methods, and the concept of automatic differentiation is applied to design a vectorised forward derivative accumulation technique; this enables provision of accurate derivatives to the SQP algorithm. The SQP serves as a local optimizer in the deterministic phase of the hybrid models. Through benchmarking on stationary and dynamic problems, results showed that the proposed models achieved sufficient diversity control, which suggests improved exploration-exploitation balance. Thirdly, to mitigate the challenges of 'inappropriate' parameter settings, this thesis proposes closed-loop adaptive mechanisms which dynamically evolve effective step sizes for the evolutionary operators. It then examines the effect of incorporating a derivative-free algorithm which extends the hybrid model to a flexible and reusable algorithmic framework.

Traditionally, solid axle railway wheelsets are stabilised by using passive suspensions on a conventional rail vehicle, but such additional stiffness affects the pure rolling action of the wheelset around the curve. It has been theoretically proven that this design conflict between stability and curving performance can be solved by applying active control instead of conventional passive components, resulting in the reduction of the wear of the wheels and track by minimising the track shifting forces. In the active approach, the use of actuators, sensors and data processors to replace the traditional passive suspension raises the issue of the system safety in the event of a failure of the active control, which could result in the loss of stability and in more severe cases, derailment. Further on, in active control systems for railway vehicles the actuators tend to be significantly more expensive and require more additional space than sensors, and an electronic control unit. Therefore, developing an analytical redundancy-based fault tolerance technique for an actively controlled wheelset that minimises the number of actuators will clearly be more beneficial. Thus the emphasis of this research is to develop a fault-tolerant system of active control for a railway vehicle in the event of actuator malfunction in order to guarantee stability and good curving performance without using additional actuators. The key achievements of this research can be summarised as follows: • The research considers three of the most common types of actuator failure for the electro-mechanical actuators: fail-hard (FH), short circuit (SC) and open circuit (OC). The fail-hard is a failure condition when the motor shaft of the actuator becomes immovable, whereas the short circuit and open circuit are failures that occur in the electrical parts of the actuator which correspond to zero voltage and zero current in the motor respectively. • The research investigates and develops a thorough understanding of the effect of actuator faults and failure modes on the vehicle behaviour that provides the necessary foundation for the development of the proposed fault-tolerant strategy. • An effective fault detection and isolation methods for actuator faults through two different approaches is developed; the vehicle model-based approach and the actuator model-based approach. Additionally, the research takes into account the reliability and robustness of the FDI schemes in the presence of sensor failures and parameter uncertainties in the system. • The research develops the control re-configuration in order to cope with the identified failure mode of the actuator in order to maintain the vehicle stability and desired curving performance.

It is often the case that robotic systems require models, in order to successfully control themselves, and to interact with the world. Models take many forms and include kinematic models to plan motions, dynamics models to understand the interaction of forces, and models of 3D geometry to check for collisions, to name but a few. Traditionally, models are provided to the robotic system by the designers that build the system. However, for long-term autonomy it becomes important for the robot to be able to build and maintain models of itself, and of objects it might encounter. In this thesis, the argument for enabling robotic systems to autonomously build models is advanced and explored. The main contribution of this research is to show how a layered approach can be taken to building models. Thus a robot, starting with a limited amount of information, can autonomously build a number of models, including a kinematic model, which describes the robot’s body, and allows it to plan and perform future movements. Key to the incremental, autonomous approach is the use of exploratory actions. These are actions that the robot can perform in order to gain some more information, either about itself, or about an object with which it is interacting. A method is then presented whereby a robot, after being powered on, can home its joints using just vision, i.e. traditional methods such as absolute encoders, or limit switches are not required. The ability to interact with objects in order to extract information is one of the main advantages that a robotic system has over a purely passive system, when attempting to learn about or build models of objects. In light of this, the next contribution of this research is to look beyond the robot’s body and to present methods with which a robot can autonomously build models of objects in the world around it. The first class of objects examined are flat pack cardboard boxes, a class of articulated objects with a number of interesting properties. It is shown how exploratory actions can be used to build a model of a flat pack cardboard box and to locate any hinges the box may have. Specifically, it is shown how when interacting with an object, a robot can combine haptic feedback from force sensors, with visual feedback from a camera to get more information from an object than would be possible using just a single sensor modality. The final contribution of this research is to present a series of exploratory actions for a robotic text reading system that allow text to be found and read from an object. The text reading system highlights how models of objects can take many forms, from a representation of their physical extents, to the text that is written on them.