My synthetic wingman must understand me: modelling intent for future manned–unmanned teaming

With advances in artificial intelligence, machine learning, and cognitive modelling, unmanned aircraft are expected to actas human-like wingmen in the near future. For fluent and effective manned–unmanned teaming, synthetic wingmen must beable to account for and adapt to their partners’ intent with little or no communication. To enable such abilities, it becomescrucial to identify the requirements that makes intent explainable to synthetic wingmen, necessitating approaches to describeand analyse intent from a human-centric perspective. To address this issue, this paper reports on findings from using WorkDomain Analysis to design and analyse models of situated intent with six levels of cognitive control (frames, effects, values,generic, implementations, and physical). Through a literature review and seven subject matter expert interviews, a synthesizedmodel was designed to represent fighter pilots’ intent in a manned–unmanned teaming scenario. Using the synthesized modelas the context, a transfer of control and a link loss situation were further described and analysed. Experiences show that WorkDomain Analysis can provide a practical and applicable means to model situated intent, particularly since designed modelscan be re-utilised to model intent in similar situations. Furthermore, the model analyses show the importance of accountingfor fighter pilots’ adopted frames since small variations of the framing of the situations can propagate throughout the modelresulting in conflicting or inconsistent intent. The paper concludes that synthetic wingmen must be able to reason about all six levels of cognitive control, requiring a more holistic approach to make intent explainable.
Funding: VINNOVA; Swedish Defence Material Administration [2017-04884,2017-04884,2017-04884,2017-04884]; NFFP (National Aviation Research Programme) - VINNOVA (Swedish Governmental Agency for Innovation Systems); Swedish Armed Forces
The Swedish Defence Material Administration and NFFP (National Aviation Research Pro- gramme), which is funded by VINNOVA (Swedish Governmental Agency for Innovation Systems, 2017-04884,2017-04884,2017- 04884,2017-04884), the Swedish Armed Forces, and the Swedish Defence Material Administration.