SUBMITTED DRAFT- HFES 2010 Impact of Conflict Avoidance Responsibility Allocation on Pilot Workload in a Distributed Air Traffic Management System Sarah V. Ligda1, Arik-Quang V. Dao1, Thomas Z. Strybel2, Kim-Phuong Vu2, Vernol Battiste1, Walter W. Johnson3 1SJSURF/NASA Ames Research Center, MS 262-2, Moffett Field, CA 94035, USA 2California State University Long Beach, CA 90840, USA 3NASA Ames Research Center, CA 94035, USA Pilot workload was examined during simulated flights requiring flight deck-based merging and spacing while avoiding weather. Pilots used flight deck tools to avoid convective weather and space behind a lead aircraft during an arrival into Louisville International airport. The study examined concepts placing conflict avoidance responsibility on different combinations of pilot, controller and automation. An ATWIT metric, modified to measure workload, showed highest workload during the approach, and lowest during the en-route phases of flight (before deviating for weather). The trend across multiple workload metrics showed workload to be lowest when pilots had both conflict alerting and responsibility for avoiding conflicts; while all objective and subjective measures showed workload was highest when pilots had no conflict alerting or responsibility for avoiding conflicts. The findings suggest workload is lowest when pilots have conflict alerting displays plus responsibility for conflict resolution; and highest when they had neither the displays nor responsibility. This suggests workload is primarily driven by an attempt to regain situation awareness when conflict alerting is unavailable. It also suggests that workload is not tied primarily to responsibility for resolving conflicts, but to maintaining situation awareness. In general, the modified ATWIT was shown to be a valid and reliable workload measure, providing more detailed information than post-run subjective workload metrics. INTRODUCTION It is predicted that demand for air travel will double within the next 15 years. To meet this demand, significant changes to the current air traffic management (ATM) system are being evaluated (Joint Planning and Development Office, 2007). It is known that increased traffic loads negatively affect air traffic controller (ATC) performance; however, this impact can be reduced when ATCs are assisted by automated conflict resolution tools (Prevot et al., 2009). One way to reduce ATC workload is to adjust the roles and responsibilities of operators in the ATM system. For example, a portion of the responsibility for maintaining safe separation distances between aircraft could be transferred from ATC to the flight deck or be automated to some extent. These strategies should alleviate a portion of ATC workload so that traffic loads can be increased while meeting or exceeding current safety and efficiency standards. Currently, both air and ground-side conflict detection algorithms have been developed to aid the human operator with identifying air traffic conflicts. These conflict detection algorithms feed data to programs such as the Auto-Resolver which can then provide conflict resolutions upon request from the user (Ezerberger, 2006). Alternatively, the Auto-Resolver can be configured to automatically provide resolutions upon detection of a conflict and wait for confirmation from ATC before executing, shifting the controller to a more supervisory position than current day operations. In the most extreme case, the Auto-Resolver can be configured to automatically generate resolutions upon detection, as well as automatically execute resolutions without prior consent from ATC. Similarly, flight deck conflict detection algorithms contain logic to identify and highlight conflicts and provide automated resolutions on displays such as NASA Flight Deck Display Research Laboratory’s (FDDRL) Cockpit Situation Display (CSD) (Granada, Dao, Wong, Johnson, & Battiste, 2005). Impact of Workload Subjective operator workload can be defined as “the perceived relationship between the amount of mental processing capability or resources and the amount required by the task” (Hart & Staveland, 1988). Many task factors can affect ATC workload, with a major contributor being air traffic density. According to Lee (2005), the relationship between workload and traffic count is non-linear. ATC workload increases from low to high only during which a certain traffic threshold is reached, meaning that workload cannot be predicted from traffic counts. Traffic management, which includes the detection and resolution of potential traffic conflicts, increases threefold with a linear increase in the number of aircraft (Wickens, 1992). This suggests that controllers reach their maximum workload capacity at a fixed traffic load, likely to be exceeded in the next 15 years. There are many strategies for automating conflict detection and resolution tasks, and optimal selection of a strategy requires assessment of workload and situation awareness. In cases where automation is completely responsible for conflict detection and resolution, humans may be thrown out-of-the-loop leading to complacencySUBMITTED DRAFT- HFES 2010 and loss of situation awareness (Parasuraman, Sheridan, & Wickens, 2000). Moreover, when workload is high, the operator is forced to tunnel in on the primary task, reducing the cognitive resources required for proactive acquisition of situation-awareness-relevant information in the environment (Parasuraman & Wickens, 2008). On the other hand, automated tools that aid the human operator might support a performance benefit because workload could be reduced without losing situation awareness (Dao et al., 2009). However, operators must trust automation to be willing to use an automated system. Studies demonstrate that this is possible with some degree of a human-in-the-loop structure and with practice and continued use (Ligda, Johnson, Latcher & Johnson, 2009). Therefore, in the current investigation, pilot workload was measured under three function allocation strategies (Concepts) for conflict detection and resolution: pilot, controller, or automation primarily responsible for conflict avoidance. This study was conducted in a larger context of trajectory oriented operations, but the present paper will focus primarily on the impact of workload for pilots in each operational concept. Measuring Workload A modified Air Traffic Workload Input Technique (ATWIT; Stein, 1985) was used in the