Comparison of Pilot and Automation Generated Lateral Conflict Resolutions

Abstract

This study compares and contrasts lateral conflict resolutions generated by pilots (with and without a set of decision support tools), with those generated by a fully automated conflict resolution tool that generates optimal (smallest path deviation) resolutions. The conflict geometries investigated were all factorial combinations of three levels of Intruder aircraft speed, three levels of initial Ownship distance to closest approach, and nine conflict angles. The resolution decision support tools included dynamic conflict alerting, which indicated whether a proposed path was conflict free, and a dynamic predictor system that showed a fast time depiction of the proposed resolution trajectories. The automation-generated resolutions, computed using a geometric optimization algorithm, served as a benchmark against which the pilot-generated resolutions were compared. Without decision support tools the pilot-generated resolutions were often ineffective in providing the necessary separation, particularly at smaller conflict angles. The resolutions tended to be effective when the decision support tools were used. Resolution cost, as measured by added path length, was greater for pilot-generated resolutions (averaging 2.7 nmi) compared to the automation-generated resolutions (averaging 1.3 nmi). When pilots had the decision support tools, their strategies, as indexed by whether they turned toward or away from the Intruder, and the mean locations at which they turned back to recapture their original route, tended to be the same as that of the automated system.