Recent accidents and incidents shave shown that operators in charge of supervisory control tasks very often have problems to maintain sufficient situation awareness (SA). A typical example is the aviation domain, where most accidents can be attributed to insufficient SA of pilots. Specific assistance systems, such as alarm systems, aim at supporting operators to maintain SA. These assistance systems observe the environment state and decide whether operators need support given a certain situation. However, the operator state (e.g. motor and glance activities) is rarely considered within the decision making process applied by these assistants. Thus, the assistance is sometimes inadequate, because the actual needs of operators are unknown. Smarter ways of assistance could be investigated if continuous assessment of operator SA during task performance would be possible. This requires a technique for assessment of operator SA, which fulfils a couple of requirements. In this paper, we present a technique for automatic assessment of operator SA, which fulfils these requirements and thus allows the investigation of new ways of assisting operators in maintenance of SA in the future. In addition, we present results of first experiments conducted with operators in charge of supervising a swarm of highly autonomous unmanned aerial vehicles. Because this paper describes work in progress, these results are not yet statistically significant. However, our results give an impression of the potential power of this technique.
08 / 2012
D3CoS Designing Dynamic Distributed Cooperative Systems