Camera field of view is far narrower than human vision

What it is

The camera field of view commonly used in teleoperation is far narrower than natural human vision — binocular human vision spans nearly 200 degrees with a clear division of labor between central and peripheral vision, while a typical robot camera covers only about 60 to 90 degrees. This limitation, sometimes called the soda-straw effect or keyhole effect, is the most concrete and measurable piece of the situation-awareness problem in teleoperation: an operator can be looking straight at a target in perfect focus while having no idea what is happening just outside the frame.

Why it happens

The cost of a narrow field of view is not just "seeing less" — it is losing the function that peripheral vision performs in natural perception. Peripheral vision has low resolution but is highly sensitive to motion and sudden change, acting as an early-warning system that tells central vision where to look next. Once a camera's field of view is compressed down to roughly what human central vision covers, the operator loses that warning layer entirely and must actively pan the camera to "scan" for changes outside the frame, whereas peripheral vision in natural viewing works passively and continuously. This gap between active scanning and passive warning is why operators routinely miss important events at the edge of the frame — an obstacle closing in from behind, a person appearing to the side. The narrower the view, the larger the angular range that active scanning has to cover, and the longer the blind window between scans. Under high workload this interacts with a further, cognitive form of narrowing — cognitive tunneling — where an operator absorbed in the primary task further contracts attention toward the center of the frame. When a physically narrow field of view compounds with this psychological narrowing, the real-world miss rate for peripheral events often exceeds what camera specifications alone would predict.

Studying it

A common design manipulates camera field-of-view angle (30, 60, 90 degrees, or a 180-degree panorama) as an independent variable, measuring reaction time to peripheral events, miss rate, subjective situation-awareness ratings, and task completion performance. Other work compares single-camera pan-tilt scanning against multi-camera stitched panoramas at matched coverage, testing whether "same coverage, different presentation" still produces a performance gap. Studies here typically also report field-of-view angle together with screen size, resolution, and viewing distance, because the same angular field of view carries very different effective information density on a small, distant screen than on a large, close one. A signal-detection framework is often borrowed to decompose the miss rate: the same drop in overall performance can come from a lower hit rate or a higher false-alarm rate, and the two point to different interface fixes.

Where it stops holding

A wider field of view is not monotonically better — excessive width introduces image distortion, reduced discriminable detail, and higher cognitive load, so field-of-view angle trades off against task demands rather than improving performance in a straight line. This limitation matters little in structured, predictable environments such as fixed rails or preset paths, and matters most in open environments where operators must stay alert to sudden events. Fisheye-style wide-angle correction is not a lossless fix either — correction crops the edges of the frame or distorts distance perception, so a geometrically corrected image cannot be assumed equivalent to a complete one. Multi-camera stitching has its own boundary too: registration error and exposure mismatch at the seams create visible artifacts that operators can mistake for real edges in the environment, and the more cameras are stitched together, the more of these failure points appear.

Applying it

Prioritize multi-camera stitching or fast pan-tilt scanning for tasks that need peripheral vigilance — outdoor navigation, work in crowded environments — instead of chasing higher resolution on a single camera. Keep a secondary view showing pose, heading, and a peripheral overview alongside the primary feed to cover angles the main image cannot. Validate by comparing, in the target task setting, how often operators miss peripheral events under different field-of-view configurations, and pick the configuration with the best trade-off between miss rate and cognitive load rather than defaulting to the technically most advanced camera.

Related

• Same group: X7.01.1 Teleoperators lack the sensory information available on site · X7.01.3 Situation awareness gaps require a supplementary environment representation
• Nearby: X7.03 Force and haptic feedback · X4.07 Operator situation awareness
• Search terms: field of view · keyhole effect · teleoperation · peripheral vision