Study of visualization modalities on industrial robot teleoperation for inspection in a virtual co-existence space

Abstract

Effective teleoperation visualization is crucial but challenging for tasks like remote inspection. This study proposes a VR-based teleoperation framework featuring a ‘Virtual Co-Existence Space’ and systematically investigates visualization modalities within it. We compared four interfaces (2D camera feed, 3D point cloud, combined 2D3D, and Augmented Virtuality-AV) for controlling an industrial robot. Twenty-four participants performed inspection tasks while performance (time, collisions, accuracy, photos) and cognitive load (NASA-TLX, pupillometry) were measured. Results revealed distinct trade-offs: 3D imposed the highest cognitive load but enabled precise navigation (low collisions). 2D3D offered the lowest load and highest user comfort but slightly reduced distance accuracy. AV suffered significantly higher collision rates and participant feedback usability issues. 2D showed low physiological load but high subjective effort. No significant differences were found for completion time, distance accuracy, or photo quality. In conclusion, no visualization modality proved universally superior within the proposed framework. The optimal choice is balancing task priorities like navigation safety versus user workload. Hybrid 2D3D shows promise for minimizing load, while AV requires substantial usability refinement for safe deployment.