No description available.

Development of manipulators that interact closely with humans has been a focus of research in fields such as robot-assisted surgery and haptic interfaces for many years. Recent introduction of powered surgical-assistant devices into the operating theatre has meant that robot manipulators have been required to interact with both patients and surgeons. Most of these manipulators are modified industrial robots. However, the use of high-powered mechanisms in the operating theatre could compromise safety of the patient, surgeon, and operating room staff. As a solution to the safety problem, the use of actively restrained passive arms has been proposed. Clutches or brakes at each joint are used to restrict the motion of the end-effector to restrain it to a pre-defined region or path. However, these devices have only had limited success in following pre-defined paths under human guidance. In this research, three major limitations of existing passive devices actively restrained are addressed.

In this thesis is presented a novel approach to model, control, and planning the motion of a nonholonomic wheeled mobile robot that applies stable pushes and pulls to a nonholonomic cart (York mail trolley) in a loosely structured environment. The method is based on grasping and ungrasping the nonholonomic cart, as a result, the robot changes its kinematics properties. In consequence, two robot configurations are produced by the task of grasping and ungrasping the load, they are: the single-robot configuration and the robot-trolley configuration. Furthermore, in order to comply with the general planar motion law of rigid bodies and the kinematic constraints imposed by the robot wheels for each configuration, the robot has been provided with two motorized steerable wheels in order to have a flexible platform able to adapt to these restrictions.

The work described in this thesis contributes to the efficient solution of mobile robot navigation problems. A series of new evolutionary approaches is presented. Two novel evolutionary planners have been developed that reduce the computational overhead in generating plans of mobile robot movements. In comparison with the best-performing evolutionary scheme reported in the literature, the first of the planners significantly reduces the plan calculation time in static environments. The second planner was able to generate avoidance strategies in response to unexpected events arising from the presence of moving obstacles. To overcome limitations in responsiveness and the unrealistic assumptions regarding a priori knowledge that are inherent in planner-based and a vigation systems, subsequent work concentrated on hybrid approaches. These included a reactive component to identify rapidly and autonomously environmental features that were represented by a small number of critical waypoints. Not only is memory usage dramatically reduced by such a simplified representation, but also the calculation time to determine new plans is significantly reduced. Further significant enhancements of this work were firstly, dynamic avoidance to limit the likelihood of potential collisions with moving obstacles and secondly, exploration to identify statistically the dynamic characteristics of the environment. Finally, by retaining more extensive environmental knowledge gained during previous navigation activities, the capability of the hybrid navigation system was enhanced to allow planning to be performed for any start point and goal point.

To achieve fully autonomous systems, fault tolerance is often employed. Fault tolerance is the ability to continue operation in the presence of faults. Fault diagnosis is an essential component of fault tolerance, especially for autonomous robotics. It is the process of determining as much information as possible about the fault, especially the origin of the fault. However, a real time fault diagnosis for resource limited robotic systems has proposed a new set of challenges, such as its complexity and efficiency, which traditional methods will find difficult to meet. This has led the work to seek inspiration from the immune system, where an effective and efficient fault diagnosis solution has been provided for thousands of years. This thesis presents a novel immune-inspired on-line fault diagnosis algorithm for robotic systems and includes the first application of that Artificial Immune System to robot fault diagnosis.

This thesis examines the problem of adaptive error detection in swarm robotics. As part of the challenges for the transition of current swarm robotics research into the real world implementation, the ability to differentiate between changes to the behaviour due to faulty components and environmental is important. This is a requirement to ensure that robot swarms deployed are fault-tolerant to internal faults as well as external perturbations. Previous work has investigated this issue from a perspective of a single robot but has largely ignored the aspect of adaptivity to environmental changes. By contrast, this work approaches the problem from a perspective of a collective and explicitly addresses the issue of adaptive detection. A collective self-detection scheme called the CoDe scheme is proposed and developed. This scheme is demonstrated to work in detecting errors in dynamic environments with the use of various classifiers. This approach has potential to be applied for other domains that share similar characteristics to swarm robotics in which adaptivity to dynamic environments is crucial. Motivated by the potential resource limitations in swarm robotic systems, this thesis also investigates other aspects related to minimising resource usage such as reducing the number of false positives and communication overhead.

In the field of human-robot interaction, socially interactive robots are often equipped with the ability to detect the affective states of users, the ability to express emotions through the use of synthetic facial expressions, speech and textual content, and the ability for imitating and social learning. Past work on creating robots that can make convincing emotional expressions has concentrated on the quality of those expressions, and on assessing people’s ability to recognize them. Previous recognition studies presented the facial expressions of the robots in neutral contexts, without any strong emotional valence (e.g., emotionally valenced music or video). It is therefore worth empirically exploring whether observers’ judgments of the facial cues of a robot would be affected by a surrounding emotional context. This thesis takes its inspiration from the contextual effects found on the interpretation of the expressions on human faces and computer avatars, and looks at the extent to which they also apply to the interpretation of the facial expressions of a mechanical robot head. The kinds of contexts that affect the recognition of robot emotional expressions, the circumstances under which such contextual effects occur, and the relationship between emotions and the surrounding situation, are observed and analyzed in a series of 11 experiments. In these experiments, the FACS (Facial Action Coding System) (Ekman and Friesen, 2002) was applied to set up the parameters of the servos to make the robot head produce sequences of facial expressions. Four different emotional surrounding or preceding contexts were used (i.e., recorded BBC News pieces, selected affective pictures, classical music pieces and film clips). This thesis provides evidence that observers’ judgments about the facial expressions of a robot can be affected by a surrounding emotional context. From a psychological perspective, the contextual effects found on the robotic facial expressions based on the FACS, indirectly support the claims that human emotions are both biologically based and socially constructed. From a robotics perspective, it is argued that the results obtained from the analyses will be useful for guiding researchers to enhance the expressive skills of emotional robots in a surrounding emotional context. This thesis also analyzes the possible factors contributing to the contextual effects found in the original 11 experiments. Some future work, including four new experiments (a preliminary experiment designed to identify appropriate contextual materials and three further experiments in which factors likely to affect a context effect are controlled one by one) is also proposed in this thesis.

As robots become more accessible to humans, more intuitive and human-friendly ways of programming them with interactive and group-aware behaviours are needed. This thesis addresses the gap between Learning by Demonstration and Multi-robot systems. In particular, this thesis tackles the fundamental problem of learning multi-robot cooperative behaviour from concurrent multi-teacher demonstrations, problem which had not been addressed prior to this work. The core contribution of this thesis is the design and implementation of a novel, multi- layered framework for multi-robot learning from simultaneous demonstrations, capable of deriving control policies at two different levels of abstraction. The lower level learns models of joint-actions at trajectory level, adapting such models to new scenarios via feature mapping. The higher level extracts the structure of cooperative tasks at symbolic level, generating a sequence of robot actions composing multi-robot plans. To the best of the author's knowledge, the proposed framework is the first Learning by Demonstration system to enable multiple human demonstrators to simultaneously teach group behaviour to multiple robots learners. A series of experimental tests were conducted using real robots in a real human workspace environment. The results obtained from a comprehensive comparison confirm the appli- cability of the joint-action model adaptation method utilised. What is more, the results of several trials provide evidence that the proposed framework effectively extracts rea- sonable multi-robot plans from demonstrations. In addition, a case study of the impact of human communication when using the proposed framework was conducted, suggesting no evidence that communication affects the time to completion of a task, but may have a positive effect on the extraction multi-robot plans. Furthermore, a multifaceted user study was conducted to analyse the aspects of user workload and focus of attention, as well as to evaluate the usability of the teleoperation system, highlighting which parts were necessary to be improved.

Existing work in the field of monocular Simultaneous Localisation and Mapping (SLAM) has largely centred around sparse feature-based representations of the world. By tracking salient image patches across many frames of video, both the positions of the features and the motion of the camera can be inferred live. Within the visual SLAM community, there has been a focus on both increasing the number of features that can be tracked across an image and efficiently managing and adjusting this map of features in order to improve camera trajectory and feature location accuracy. Although prior research has looked at augmenting this map with more sophisticated features such as edgelets or planar patches, no incremental real-time system has yet made use of every pixel in the image to maximise camera trajectory estimation accuracy. Moreover, across many practical domains, these feature-based representations of the world fall short. In robotics, sparse feature-based models do not allow a robot to reason about free space and are not so useful for interaction. In augmented reality, sparse models do not allow us to place virtual objects behind real-ones and cannot enable virtual characters to interact with real objects. In this research we show how a dense surface model offers many advantages and we explore different methods of reasoning about dense surfaces over a sparse feature-based map. We continue by developing different methods for dense tracking and constrained dense SLAM in different applications such as spherical mosaicing. Finally, we show how live dense tracking can be tightly integrated with dense reconstruction to create a 6 DOF monocular live dense SLAM system which outperforms the current state of the art in many respects.

No description available.