Quantity or quality - predictability and experience: A case study in human-robot interaction

Abstract

Human-robot collaboration is essential when the robot operates in unstructured environments which change dynamically and require high perception capabilities. The design of such collaborative systems requires the system to be predictable, i.e., the system's response should be repeatable and as close as possible to the response expected by its users. This should enable users to comprehend and learn how the system operates and foresee the system's responses to their commands and actions. We present a study of remote controlling a robot. This study investigated the expectations, perception, behavior and preferences of users while issuing a "forward" movement command. The study aimed to determine if users expect and prefer a system response that is identical in the distance of movement (repeated quantity), or an adaptive movement whose size depends on the environment (i.e., repeated rule, such as movement until another command, until junctions and/or obstacles). Speech was the only modality of interaction in the two experiments performed. The results show that the movement manner adaptable to the environment is preferred. Although the manner of movement (set step size vs. continuous movement, stopping at junctions vs. not stopping) may affect the overall performance, especially in the learning stage, these differences are not always perceived by the users. Results indicate that a robot's response could be qualitatively similar rather than identical in quantity or quality (the direction is constant, but the movement and feedback manners may vary). Furthermore, the overall gained user experience compensates for minor variations in the system's response.