Carl Westin

3-2 Automation acceptance in ATC 41 controller might opt to either disuse (i.e., make no use of) or misuse (i.e., use it in a manner inconsistent with design intents) the new automation. Several efforts to introduce advanced CD&R automation have been thwarted over the years specifically because of acceptance problems. 10, 12, 77 Acceptance problems have been linked to such interrelated issues as: • automation misuse; 9, 13, 130 • perceived strategy mismatches between controller and machine, 31 and • system performance decrements, a result of the controller having to “fight” the system. 17 Given that controller initial acceptance appears to have impeded some past ef- forts to develop advanced CD&R automation, 14 and given the evolutionary trajec- tory that such automation appears to be following (toward more strategic capabili- ties), it is reasonable to hypothesize that controller acceptance will present at least as big an issue in the future. We therefore sought to explore how and whether ac- ceptance could be impacted by the closeness of fit between human and machine problem solving styles. 3-2-1 Strategic conformance between controller and automation Several authors have noted potential benefits of heuristic, as opposed to algorith- mic, approaches to CD&R automation. 26, 29, 30, 32 Notice here the distinction: a so- lution might be operationally or mathematically optimized (think of a system that always chooses the shortest route), but this solution is not necessarily the one that the controller would select. 19, 72 Controllers’ preferred solutions might be driven by, for instance, general control strategies, 89 (e.g., guaranteeing separation through course divergence) or even display perceptual limitations (e.g., a tendency to disre- gard speed control in en route airspace). There have been various attempts to explicitly model the controller’s conflict resolution strategies, 31, 35, 79, 131, 132 although efforts to embed these in automation have proven difficult. 19, 133 One such effort is EUROCONTROL’s CORA, which aimed to create a conflict resolution decision aid around a core of controller heuristic solutions. 35 In the initial phase, controllers would be presented an ordered list of candidate solutions (based on previous group responses) from which the controller could choose. 3-2-2 A model of controller automation usage The MUFASA project laid out a functional model of controllers’ automation ac- ceptance, as shown in Figure 3-1. 134 In this simplified model, various internal and