Evaluating Camera Viewpoint Control in Telemanipulation: Performance, Workload, Visual Attention, and Usability

Camera configuration plays a critical role in telemanipulation, where operators rely on video feeds to perceive and interact with remote environments. However, selecting the number, placement, and control mode of cameras poses challenges in balancing visual coverage, cognitive workload, and task efficiency. This study investigates how different camera viewpoint control models affect operator performance and workload in telemanipulation tasks. A user-centered study with a repeated measures design experiment was implemented. The telemanipulation scenario was based on a cube-stacking task. Five camera viewpoint control models were studied: (1) three fixed cameras, (2) two fixed cameras and a dynamic camera with deep-learning-based autonomous viewpoint adjustment, (3) two fixed cameras and a dynamic camera with manual viewpoint adjustment, (4) a single dynamic camera with autonomous viewpoint adjustment, and (5) a single dynamic camera with manual viewpoint adjustment. Study outcomes were operational performance (i.e., cube-stacking success and time spent on stacking), cognitive load (i.e., eye blink rate, pupillary activity, and perceived workload), visual attention (i.e., eye fixation rate and saccade rate), and perceived usability. A generalized linear mixed-effects model was used to examine the effect of camera viewpoint control models on cube-stacking success. Differences in other outcomes across the five control models were analyzed using one-way repeated measures analysis of variance.
Our results show that multiple-camera systems improve task success but introduce greater cognitive and visual demand. Dynamic-camera models, particularly those with autonomous control, enhance usability and may support more efficient and less visually taxing interaction. These findings have direct implications for the design of telemanipulation interfaces in high-stakes domains such as robotic surgery, industrial automation, and remote maintenance, where camera control significantly impacts task performance and user well-being.