学术文献库

Some applications require an assurance that certain criteria are violated with only low probability. An alert is generated when the current course of action is likely to violate assurance criteria and the alert results in corrective action. Assurance monitors fuse information from multiple data streams generated by sensors and other sources to estimate the probability distribution of system trajectories. This distribution is used to determine whether an assurance constraint is likely to be violated. The acquisition of data requires resources such as energy, computational power and communication bandwidth which are scarce in many applications. At each point in time the system must decide whether to expend these resources to get more data to improve the confidence in the probability distribution of trajectories. This paper presents a software-design pattern for assurance monitoring and gives examples of its uses, including an application of the pattern to the problem of autonomous navigation of a drone which is required to avoid no-fly zones while using limited resources.