Human–Robot Collaboration: Understanding the Effects of Different Levels of Automation Assistance on Enhancing Eye-Gaze-Based Mobile Robot Tele-operation

The introduction of autonomous technologies in cities—such as driverless cars, ground robots, and drones—promises to transform urban services and interactions. As a newly emergent, intuitive, and promising technology, eye-gaze-based mobile robot teleoperation (referred to as Tele-gaze in this context) has recently been applied to human–mobile robot interaction for various activities in cities, such as in the use of assistive technologies to improve the autonomy and well-being of people with limited mobility. More recently, we have begun to witness a shift that takes advantage of automation assistance to improve human performance using Tele-gaze. However, there is a knowledge gap with regard to how various levels of automation assistance, when combined with interaction modalities, can affect human operators and their collaboration with non-human agents. More specifically, we need to identify the factors that can influence the Tele-gaze-based systems and their effects on the human–robot collaboration. In this paper, a novel Tele-gaze system supported by four interactive modalities is proposed. The eye-gaze control is supported by increasing levels of automation assistance from the mobile robot side, with two basic mechanisms (eye-imitation camera and user-will trigger mechanisms) present in all four modalities—Modality 1: without any automation assistance; Modality 2: with a collision-avoidance mechanism; Modality 3: with a collision-avoidance mechanism and pathway middle mechanism; and Modality 4: with a collision-avoidance mechanism and a guideline-following mechanism. Four sets of experiments with an indoor navigation task were conducted with the four interactive modalities. Each set was tested by our 12 participants using one of the four modalities. We analyzed and discussed the experiments’ outcomes from both (1) the mobile robot side and (2) the human operator side. Early findings indicated that automation assistance and human capacity can not only reinforce one another but also constrain one another. We suggest that adopting the right balance, with the appropriate level of automation assistance, is key to enhancing the Tele-gaze-based interactions without compromising human dominance, user experience, and the flexibility of the system capacity.