Toward Reliable Closed-Loop Cybersickness Mitigation in Head-Mounted Virtual Reality

Abstract

Cybersickness remains a barrier to sustained use of head-mounted display (HMD) virtual reality (VR) and is often evaluated as a post-exposure outcome rather than as a changing user state requiring intervention during immersion. This focused narrative review used structured searches of Google Scholar, Scopus, and Web of Science to examine evidence for sensing, state estimation, mitigation, reassessment, and adaptive control; 47 sources were included in the final synthesis. Physiological, ocular, kinematic, behavioral, and content-derived signals can support estimation of cybersickness-related outcomes, but their usefulness depends on the reference measure, temporal alignment, validation design, and operating context. Visual, locomotion, rendering, and multisensory interventions can reduce symptoms, but no strategy is consistently superior, and mitigation can impose costs to visual access, usability, or task performance. Existing architectures demonstrate connected estimation and adaptation, but evidence remains limited for prospective reassessment, de-escalation, responses to failure to improve, generalization, and safety supervision. The review organizes these findings within a framework linking dynamic user state, sensing, preprocessing, actionable state estimation, intervention selection, system adaptation, reassessment, and updated user state. Reliable closed-loop mitigation requires prospective validation of the feedback cycle, including total latency, control-error consequences, user override, limits on intervention magnitude and adjustment rate, and exposure-termination criteria.