Operator visual attention allocation prediction in a robotic arm teleoperation interface

Abstract

In digital interactive interfaces with high visual workloads, it is important for operators to allocate their limited attentional resources appropriately to ensure efficient information collection. The salience, effort, expectancy, value (SEEV) model, which combines top-down and bottom-up attention mechanisms for predicting attention allocation, has been validated in research areas such as piloting, driving, and surgical operations. However, the validity of the SEEV model in the field of robotic arm teleoperation has not yet been thoroughly studied. The primary purpose of this study was to confirm the feasibility of the SEEV model for operator visual attention allocation prediction in a robotic arm teleoperation scenario. The improved ITTI algorithm, distance-measuring tool, Delphi method, and lowest ordinal algorithm were adopted to qualify the four factors of the SEEV model, which also contributed to salience and expectancy quantification methods. Accordingly, an attention allocation prediction model in a robotic arm teleoperation scene was constructed. To verify the validity of the prediction model, 20 participants were recruited to control the robotic arm using V-REP simulation software, and their fixation durations were recorded using an eye tracker as an attention allocation indicator. Participants controlled the robotic arm according to the experimental requirements and operational tasks, such as grasping and placing the target. The results demonstrated that the theoretical data based on the SEEV prediction model are significantly related to the proportion of fixation durations. The experiment verifies the suitability of the SEEV prediction model, and it is anticipated to be utilized in the optimization of interactive interfaces for robotic arm teleoperation.