Model of autonomous fault tolerance Sample Clauses

Model of autonomous fault tolerance. ‌ The robust autonomy is a non-functional requirement and as such does not have well-defined specifications or metrics. The foremost step in the development towards robustness for au- tonomous robots is to determine the concepts that can be used for optimisation of design so- lutions. The amount of control that a robot has over its actions, reflects different levels of autonomy [13]. The ability to prioritise and make decisions is a property of a system with a high-level of autonomy. This ability becomes more apparent in the case of abnormal events. Exceptional situations unforeseen by the main operation workflow, confront a robot with the challenge of choice. For example, an automatic ground collision avoidance system (auto GCAS) on an aircraft has a priority of the crew safety, which can override a direct command from the pilot controls in case of a ground collision course [28]. In this case the automatic system exhibits the highest level of autonomy, taking over the responsibility of making decisions. The process of a robot service can be seen as a sequence of states of the system consisting of a robot and the environment. During a normal workflow transitions between different states are determined by certain pre-programmed behaviours. However, deviations due to irregularities require an unpredicted switch in the behavioural model. Such a switch might require employ- ing one of the predefined solutions or developing a new one based on the system goals and constraints. The ability to independently detect abnormal events and to recover from them makes a robot autonomous and robust. In essence, robust autonomy is the ability of a robot to deal with abnormal situations with minimal human involvement [16, 27]. The metrics of robust autonomy can be identified through the analysis of processes taking place in the system upon an event. An abnormal event is a violation of the operating assump- tions and thus can be treated as an external fault on the boundary of the system consisting of the robot components. Following Xxxxxxxxx, [2] taxonomy of Dependable and Secure Comput- ing, the fault handling can be modelled by two processes: fault detection and recovery from it. Metrics of robust autonomy can be defined as a combination of characteristics of the fault detection and recovery processes. The field of service robotics is based on the idea of robots co-existing with humans; therefore, the human-machine interaction aspect of robust autonomy should be quantified ...