Evelyn Wilson¹, Quan Lei²; ¹Wichita State University, ²Wichita State University

Automation has been widely used in all sectors of human activities yet how automation affects basic visual processing involved in human performance is understudied. The current study examined how the level of automation and the takeover lead time when automation fails affect time-to-collision (TTC) judgments in a simplified scenario simulating human-automation interaction. Participants (n=17) watched a series of two objects (2°) moving horizontally on a screen that either collided or passed each other. In the high automation condition, automation was implemented on 2/3 of the trials where participants passively watched the objects moving with no response required, while on the remaining 1/3 of trials, motion of the objects was occluded midway, signaling the failure of automation, and participants had to take over and make a TTC judgment. In the low automation condition, automation was implemented on 1/3 of the trials while participants had to take over on the other 2/3 of trials. There was also a no-automation condition where participants had to respond on all trials. To manipulate takeover lead time, objects moved at either a faster or a slower speed. TTC judgment accuracy was measured as the deviation of estimated TTC from the actual TTC. Unlike the typical underestimation reported in previous research involving collisions with the observer, TTC judgments in this study were overestimated overall. There was a significant effect of speed (or takeover lead time) on TTC accuracy, the judgments being less accurate with shorter lead time. However, no effect of automation level was found nor any interaction between automation level and takeover lead time, despite a non-significant trend for TTC accuracy to improve with higher automation level. Our results suggest that takeover lead time plays a major role in TTC judgment in scenarios involving human-automation interaction.