Depth Perception and Task Performance in Robotic Teleoperation

Abstract

Accuracy in depth perception is critical for robotic teleoperation. This is especially true for tasks involving grasping, relocating, manipulation, etc. This study aims to improve human operators' depth perception and task performance using 3D visual cues. The working hypothesis is that providing 360-degree (3D) video as visual cues would significantly improve depth perception compared to a traditional 2D video. This study aimed to answer two research questions: (1) Does providing a 3D video improve operator depth perception and task performance in a simulated teleoperation? (2) How does wearing a VR headset affect operator comfort and mental workload? Twenty-three human subjects were recruited to perform five teleoperation tasks using both 2D and 3D visual cues. To view the 3D visual cues, a virtual reality (VR) headset was used. On the other hand, 2D video was shown on a 50-inch TV. The tasks involved grasping objects (different sizes and shapes) from one location and placing them in another location. Results showed that compared to 2D visual cues, participants had better task performance with 3D visual cues in terms of task completion rate, task completion time, and errors made. Participants also expressed that 3D visual cues were more useful in their teleoperation tasks, and they had a lower mental workload with 3D visual cues.