Does acceptance of automated solutions differ when a controller thought that the machine had provided a solution? Do controller resolutions differ from those of automation and do they consent to automated solutions? Does consent vary by conformity (i.e. whether automated solutions differed from their own)? Do controllers reject their own decision when they mistakenly believe these come from automation?
MUFASA aims to develop a framework for designing future levels of ATM automation, based primarily on human-in-the-loop simulation. The project scope was to study the benefits of ‘strategic conformance’, i.e. the degree to which automation’s behaviour and apparent underlying operations match those of the human. MUFASA started from the assumption that future ATM will increasingly rely on automation that can assume control of the cognitive and strategic aspects of ATM. The controllers were presented with advisories which they were led to believe were generated by automated agents but which were actually replays of their own conflict resolutions in earlier simulation runs.
The project was able to show that conformal advisories were accepted more frequently, led to higher controller agreement and also reduced acceptance time. This has some potential implications for the design of automation as it seems that higher trust and acceptance can be obtained through adapting automation logic to an individual’s working style. Another interesting finding within MUFASA was that approx 25% of controllers did in fact reject their own decisions when they believed they were generated by automation (roughly one out of four cases). The reasons for this could be an inherent automation bias that could be the cause or whether controllers are inconsistent over time could be subject for future research.