Extreme Environments and Extreme Science: Reliability and Risk Assessment for Autonomous Systems with Application to Polar Campaigns

Assessment of reliability and consequent risk to autonomous systems is an increasingly common and critical concern given the number of challenging new extreme environment research programs calling for the utilization of autonomous systems (e.g. AUVs, gliders, floats, etc.). The interest in using autonomous systems amongst the scientific community is particularly strong amongst polar research programs where so much vital area lies beyond the reach of traditional approaches. Therefore, the potential for scientific discovery is significantly increased, because of the very ability of autonomous systems to get to and gather information in the critical zones. The scientific merits and rewards of polar research are great but so too are the risks. There are risks both to mission success (i.e. science delivery) and risks to asset survival and recovery (i.e. retrieval). In polar settings the greatest increased risks are the complexities of operation (e.g. launch/recovery and retrieval) associated with sea ice and shelf ice. Even in open water settings the temporal and spatial dynamics of environmental conditions complicate the operation of autonomous systems. Very little systematic study and quantitative analysis has been conducted to evaluate the reliability and risk to autonomous systems in any operational setting let alone the demanding and increasingly sought after polar environments. Here we present some assessments of AUV reliability for polar and also non-polar settings drawing largely on datasets from both a large autonomous vehicle program (Autosub) and a small autonomous vehicle program (DOERRI) in order to illustrate key and common elements of reliability and risk that may provide insights to scientific end-users (PIs), program managers, and the developers and operators of other similar autonomous systems working in polar settings. An approach to risk management is laid out. Key risk mitigation elements are presented in categories of system stability and operational protocol based on existing datasets and hypothetical polar campaign scenarios. In particular we explore risk and reliability from the perspective of two key stakeholders 1) the owner/technical operator and 2) the scientific PI. Resolving the tensions and mutual interests between these two parties and the specific objectives and criteria for success/failure that they each bring to the table are at the center of a risk management program that must be undertaken prior to polar operations. We hope to engage a dialogue between scientific end-users and technical operators through a case study example demonstrating a framework for risk management suitable to autonomous systems especially in polar settings.