Xu, X. s , & Rantanen, E. M. (2003). Conflict detection and resolution in air traffic control: A task analysis, literature review, and need for further research. Proceedings of the 12th International Symposium on Aviation Psychology (pp. 1289-1295). April 14-17, 2003, Dayton, OH. 1 CONFLICT DETECTION IN AIR TRAFFIC CONTROL: A TASK ANALYSIS, A LITERATURE REVIEW, AND A NEED FOR FURTHER RESEARCH Xidong Xu and Esa M. Rantanen University of Illinois, Aviation Human Factors Division Savoy, Illinois, USA A task analysis for a typical air traffic control conflict detection scenario is described. Two types of prediction components are identified, (1) a relative judgment (RJ) task with regard to which aircraft will reach the projected point of intersection first, and (2) a prediction-motion (PM) task estimating the time-to-collide if the aircraft will reach the intersection simultaneously. Also identified are several alternative strategies that can be potentially used by controllers to perform the RJ and PM tasks. Following the task analysis, a literature review on factors influencing the RJ and PM performance reveals that although there are a number of studies on controller performance in conflict detection, almost no one has studied how the controller actually performs the RJ and PM tasks. Finally, a number of potential research questions are proposed with an aim to shed new light on the human attention/cognition mechanisms involved in the prediction tasks in ATC and hopefully contribute to the knowledge base regarding motion prediction involving multiple objects or multiple tasks. Introduction Many air traffic control (ATC) job analyses consider prediction as one of the controller’s core skills. For example, Roske-Hofstrand and Murphy (1998) conclude that although different ATC positions constitute different cognitive task requirements due to the different aircraft flight phases, task aids, and time constraints for each position, almost every such position demands mental projection or estimation. Prediction is also considered an important part of situation awareness (SA), which is considered to consists of three hierarchical phases: Level 1 SA, perception of the elements in the environment, Level 2 SA, comprehension of the current situation, and Level 3 SA, projection of future status (Endsley, 1995). In the ATC context, the controller needs to predict what will happen in the future based on Levels 1 and 2 SA to achieve Level 3 SA (e.g., two aircraft on converging courses will probably collide in three minutes if no measure is taken). The task of conflict detection is very resource demanding. Indeed, midair collisions and near- misses have repeatedly occurred due to controller's inability to predict the loss of separation between aircraft (e.g., the midair collision between a Russian passenger plane and a British cargo plane over Germany in 2002). Prediction is largely dependent on spatial working memory to “compute” the future states of aircraft (Wickens, Mavor, & McGee, 1997). When relative positions of multiple aircraft need to be predicted, the controller’s processing capabilities will be heavily taxed, thus limiting the accuracy of prediction. Understanding the underlying perceptual and cognitive mechanisms is therefore necessary in order to identify the human limitations in conflict detection and to determine areas where automated aids are most beneficial. Task Analysis Although there are occasions where predictions involving more than two aircraft need to be made, we will for the sake of simplicity limit the discussion here to situations where only two aircraft are present. Consider two aircraft flying at the same altitude on straight converging courses, at constant velocities (see Figure 1). Assuming that both aircraft will maintain their current headings, airspeeds, and altitudes, they may collide at the projected intersection point on their extended trajectories. Note that for a more general case (3-D), the overlapping of two aircraft’s positions on the horizontal plane is only necessary but not sufficient for the two planes to collide; the controller needs to perform predictions for the horizontal and the vertical planes at the same time. The prediction of the latter—given vertical speeds and current altitudes of the two aircraft— involves computation whether the planes will reach the same altitude at the time of the position overlap on the horizontal plane. Further, if instead of using a point as the criterion for collision we use a 5 nm and ± 1000 ft separation as the conflict criterion, the same reasoning would still apply. In this case, we will increase a point to a volume of space as the protected zone for an aircraft into which no other aircraft is allowed to penetrate. Relative judgment and prediction-motion tasks. First, the intersection point in Figure 1 needs to be extrapolated, which is a relatively easy task for straight and constant courses. Whether and when the aircraft will collide also need to be predicted, which