Exploring the Visual Space to Improve Depth Perception in Robot Teleoperation Using Augmented Reality: The Role of Distance and Target’s Pose in Time, Success, and Certainty
Résumé
Accurate depth perception in co-located teleoperation has the potential to improve task performance in manipulation and grasping tasks. We thus explore the operator’s visual space and design visual cues using augmented reality. Our goal is to facilitate the positioning of the gripper above a target object before attempting to grasp it. The designs we propose include a virtual circle (Circle), virtual extensions (Extensions) from the gripper’s fingers, and a color matching design using a real colormap with matching colored virtual circles (Colors). We conducted an experiment to evaluate these designs and the influence of distance from the operator to the workspace and the target object’s pose. We report on time, success, and perceived certainty in a grasping task. Our results show that a shorter distance leads to higher success, faster grasping time, and higher certainty. Concerning the target object’s pose, a clear pose leads to higher success and certainty but interestingly slower task times. Regarding the design of cues, our results reveal that the simplicity of the Circle cue leads to the highest success and outperforms the most complex cue Colors also for task time, while the level of certainty seems to be depending more on the distance than the type of cue. We consider that our results can serve as an initial analysis to further explore these factors both when designing to improve depth perception and within the context of co-located teleoperation.
Origine | Fichiers produits par l'(les) auteur(s) |
---|