Locomotion as a performance variable in standalone virtual reality systems: an ISO/IEC 25010—aligned experimental study

Locomotion techniques in virtual reality are typically evaluated from a user experience perspective, while their impact on real-time system performance remains insufficiently understood. This study investigates whether locomotion paradigms—namely continuous movement, dash teleport, and teleportation—constitute measurable system-level performance factors in standalone VR. A fully crossed 3 × 4 × 3 factorial experiment manipulated locomotion technique, scene complexity, and display refresh rate across 720 controlled benchmark runs on a Meta Quest 3 standalone headset. Traversal distance and path geometry were held constant across locomotion conditions to isolate temporal workload distribution from spatial variability. Performance efficiency was evaluated using metrics aligned with ISO/IEC 25010, including frame-time statistics, CPU and GPU processing time, energy consumption, and memory behavior. Data were analyzed using linear mixed-effects models with repeated measurements. Results indicate that scene complexity and refresh rate are the dominant determinants of performance across most metrics. Locomotion technique exhibited smaller but statistically significant effects, particularly for CPU and GPU processing time and frame-time variance. These differences were most pronounced under high computational load, where continuous locomotion produced higher processing costs than teleport-based techniques. In scenes dominated by dynamic agents, locomotion-related differences were negligible, suggesting that agent simulation can overshadow locomotion-related overhead. Overall, locomotion is not performance-neutral in standalone VR. While its impact is modest relative to scene complexity and refresh rate, it influences runtime workload distribution under constrained frame-time budgets. These findings position locomotion selection as a performance-relevant design consideration, particularly when limited frame-time headroom coincides with computationally demanding workloads.