Objective: This experiment examined whether tele-operators learn to better judge a robot's ability to pass through an aperture, hereafter referred to as pass-ability judgments, and detailed the nature of such learning. Background: Jones, Johnson, and Schmidlin reported that tele-operators' pass-ability judgments did not improve over the course of their experiment, which was surprising. Method: In each of seven blocks, tele-operators made pass-ability judgments about 10 apertures whose width varied. During each trial, participants drove the robot toward the aperture, answered yes or no to whether it could pass through that aperture, and then attempted to drive the robot through the aperture. Pass-ability judgments were analyzed in terms of percentage correct and absolute thresholds; the latter mimicked how Jones et al. analyzed their data. Results: Learning was revealed when judgments were analyzed in terms of percentage correct and not when analyzed in terms of absolute thresholds. Further analyses revealed that tele-operators only improved their pass-ability judgments for impassable apertures, and tele-operators' perceptual sensitivity and response bias changed over the course of the experiment. Conclusion: The percentage correct-based analyses revealed that tele-operators learned to make better pass-ability judgments. Jones et al.'s decision to analyze their data in terms of absolute thresholds obscured learning. Application: The present results suggested that researchers should employ percentage correct when studying learning in this domain, training protocols should focus on improving tele-operators' abilities to judge the pass-ability of impassable apertures, and tele-operators truly learned to better discriminate passable and impassable apertures.
- human-robot interaction
- perceptual-motor performance