Document Type


Publication Date



This article presents a first concept of a pilot assistant system that adapts its support to the current intent of the pilot during Manned-Unmanned-Teaming (MUM-T) helicopter missions. Assistant systems often depend on a pre-defined plan. Due to unpredicted situational changes, the plan can deteriorate, and the system is not able to assist anymore. We envisage a system design that will infer the pilot’s intent by using a domain theory approach (plan recognition as planning). To compose a possible plan, a sequence of decisions about the relevant actions is necessary. Thus, we formulate sequential planning problems using Partially Observable Markov Decision Processes (POMDP). POMDP enables us to consider the uncertainty of the mission’s course and environment. To perform human-in-the-loop experiments, the next steps are to develop the functions of the designed assistant system and integrate them into our mission and cockpit simulation environment.