Uncertainty Expression for Human-Robot Task Communication

TL;DR

This work introduces the Uncertainty Expression System (UES) to enable end users to convey uncertain scene insight about task-critical objects to robots prior to runtime, addressing scenarios where object locations are unknown. It compares three interfaces—precision (exact probability distributions), paint (heat-map), and rank (ordered locations)—in terms of accuracy, efficiency, usability, and cognitive load, finding that the rank interface offers the best overall user experience while paint tends to be less accurate. A planning case study demonstrates how user-provided scene insight can guide execution, though benefits vary with task assumptions. The findings inform the design of human-in-the-loop robotic planning tools by highlighting when simple, user-friendly uncertainty expression can outperform more precise but burdensome methods.

Abstract

An underlying assumption of many existing approaches to human-robot task communication is that the robot possesses a sufficient amount of environmental domain knowledge, including the locations of task-critical objects. This assumption is unrealistic if the locations of known objects change or have not yet been discovered by the robot. In this work, our key insight is that in many scenarios, robot end users possess more scene insight than the robot and need ways to express it. Presently, there is a lack of research on how solutions for collecting end-user scene insight should be designed. We thereby created an Uncertainty Expression System (UES) to investigate how best to elicit end-user scene insight. The UES allows end users to convey their knowledge of object uncertainty using either: (1) a precision interface that allows meticulous expression of scene insight; (2) a painting interface by which users create a heat map of possible object locations; and (3) a ranking interface by which end users express object locations via an ordered list. We then conducted a user study to compare the effectiveness of these approaches based on the accuracy of scene insight conveyed to the robot, the efficiency at which end users are able to express this scene insight, and both usability and task load. Results indicate that the rank interface is more user friendly and efficient than the precision interface, and that the paint interface is the least accurate.

Figures (6)

• Figure 1: With the Uncertainty Expression System, we investigate three different interfaces that enable humans to express possible object locations to the robot. From this scene insight, the robot can create a task plan to achieve an objective.
• Figure 2: (Top) Preliminary scene information that the robot must possess. (Bottom left) scene insight from the precision and rank interfaces are calculated by determining the closest waypoints to user-selected points. (Bottom right) scene insight from the paint interface is calculated by summing the brightness of pixels closest to each waypoint.
• Figure 3: The three UES interfaces. Precision (left): users select points where objects may exist and use sliders to express their probabilities. Paint (center): users express probabilities by painting a heat map. Rank (right): users select and rank points from most to least likely. For precision and rank, points are visualized as red circles. Circles that do not fit on an area are cropped.
• Figure 4: The procedure for each trial involved four steps. First, (1) participants watched a tutorial video on a particular interface. Next, (2) participants became familiarized with ground-truth probabilities of objects being in different locations. Then, (3) participants used the interface to express scene insight to the robot. Lastly, (4) participants answered a questionnaire.
• Figure 5: The results for our subjective measures. Error bars represent standard error of the mean. Lower values are better for all measures except for SUS. *p < 0.1, **p < 0.05, ***p < 0.01