Locomotion decides comfort, and the choice is a trade rather than a best practice

Locomotion ties a spatial system's three hardest constraints together: how much presence it produces, how sick it makes the user, and how much room it needs. Its most useful result is a rule rather than a recommendation: every technique wins on one axis by giving ground on another, so choosing a method is a decision with a stated cost. This sixth of ten units follows the body into movement, where that cost is paid.

Natural locomotion buys ease and presence, and naturalness does not predict performance

Usoh and colleagues (1999) replicated an earlier walking-in-place versus flying study and added real walking as a third condition. Real walking was significantly better than both on ease and higher on presence, but the strongest result lay elsewhere: presence correlated with how far participants associated with their avatar.

Whitton and colleagues (2005) show why that is easy to read as "natural is better". Participants walked to wall targets under three locomotion interfaces and three visual conditions. Over 95 per cent of the variance in the motion paths sat in three values, and a naturalness ordering appeared in the movement but not in task performance, so naturalness predicted how the movement looked and not how well the task went.

Nabiyouni and colleagues (2015) compared real walking, a game controller, and a semi-natural rotating sphere, the Virtusphere, in which walking inside the sphere drives the viewpoint. Measured by path deviation, the sphere was significantly slower and less accurate than both of the others. The loss traces to the forces and motions the device imposes when a walk starts and stops, which differ from those of a real step. Realism and performance do not rise together.

Redirected walking fits walking into a room by breaking the mapping

Most rooms are too small to walk through a virtual world. The field therefore kept the proprioceptive feedback of walking, and broke the mapping between the physical step and the virtual one. Interrante and colleagues (2007) proposed seven league boots, which estimate the intended direction of travel and amplify only the motion aligned with it.

Redirected walking leaves the step alone and rotates the world so a straight physical path curves. Steinicke and colleagues (2010) measured how much manipulation goes unnoticed with twelve participants in three forced-choice experiments. Users could be turned about 49 per cent more or 20 per cent less than they perceived, and distances downscaled by 14 per cent or upscaled by 26 per cent. A circular arc of radius greater than 22 metres felt straight, so below the thresholds the manipulation is invisible.

Space can also be manufactured. Suma and colleagues (2012) built impossible spaces, architecture that overlaps itself so a larger interior fits a smaller tracked area. Small rooms overlapped by up to 56 per cent before detection, and rooms filling a 9.14 metre square workspace by up to 31 per cent, with distances still seeming uncompressed.

The threshold is not a constant, and the method is a choice

It is tempting to treat a threshold as a design constant, and the literature forbids it. Grechkin and colleagues (2016) re-measured the curvature threshold with translation gain also applied and found no evidence that it changed, so the two can safely be combined. But their estimates of the detectable radius were 11.6 or 6.4 metres, depending on the estimation method, against the 22 metres reported earlier.

What the world is steered around is also a choice. Hodgson and Bachmann (2013) compared four redirected walking algorithms on simulated paths and live users. They measured distance from the centre of the tracking area, wall contacts and redirection rate. Their centre-steering algorithm outperformed the other three, so two implementations can need different space for the same journey.

Redirection is not free of attention either. Bruder, Lubos and Steinicke (2015) had users perform verbal and spatial working memory tasks while walking under redirection. A manipulation the user cannot see is still a task they are performing.

Teleportation abandons continuous travel, and the trade survives the comparison

The cleanest escape is to stop moving continuously. Bozgeyikli and colleagues (2016) introduced point and teleport, in which the user aims at a destination and is moved there. They compared it with walking-in-place and a joystick over two experiments with sixteen users. With no visible translational motion it is not expected to cause sickness, and participants found it fun and user friendly, though a direction component degraded the experience.

Teleportation is what many virtual reality experiences ship. Prithul, Adhanom and Folmer (2021) describe it as the de facto locomotion method for many such experiences, reporting that it lets a user navigate beyond the tracking space with a low possibility of inducing sickness. Anderton, Creed, Sarcar and Theil (2024) review the locomotion options in 330 of the most used applications released between 2016 and 2023 and find that description dated: teleportation fell from 48 per cent of 2016 releases to 18 per cent in 2023, while climbing rose from 3 per cent to over 30 per cent.

The head-to-head comparison makes the trade visible. Langbehn, Lubos and Steinicke (2018) compared joystick, teleportation and redirected walking on spatial knowledge, effectiveness, sickness, presence and preference. Redirected walking gave the best spatial knowledge acquisition, and it and teleportation were preferred over the joystick, which caused a significant increase in sickness. No technique led on every measure: the one that preserved spatial knowledge demanded the most space.

The contested question: one best technique, or one per task?

The literature divides over whether locomotion has a best practice at all. On the natural-walking position, walking is the technique to preserve: it gave the greatest presence and ease (Usoh et al., 1999), and the manipulations that fit it into a small room can be made hard to detect, combined safely and built into the architecture (Steinicke et al., 2010; Grechkin et al., 2016; Suma et al., 2012). On this account redirected walking is the technique to beat.

On the task-contingent position, no technique holds the top of the table. Performance did not cluster by naturalness (Whitton et al., 2005). The semi-natural technique lost to the low-fidelity controller (Nabiyouni et al., 2015). Each technique led on a different measure in the head-to-head comparison (Langbehn et al., 2018). The field also lacks a shared vocabulary: a review of 2014 to 2017 work identified thirty-six articles covering eleven techniques, and found the research concentrated on technology rather than experience (Boletsis, 2017). Dylong and Nickel (2025) add a fourth axis, mental workload, comparing two variants each of point-teleport and directed-steering with twenty participants. Point-teleport with a curved ray gave a lower workload than the straight-ray variant, while both directed-steering variants impaired performance, mainly through visually induced motion sickness.

What would settle it is one within-subject study fixing the virtual task and the participants. It would compare real walking, redirected walking, teleportation and joystick on presence, sickness, spatial knowledge and workload at once, recording the space each demands. No study runs that design. The closest (Langbehn et al., 2018) omits workload, so it cannot say whether the technique that preserved spatial knowledge also cost the most attention. Until then, the honest recommendation stays conditional: state the task, the room and who has to move in it, since real and redirected walking both assume a tracked space and someone who can stand and walk. Then say which axis you are buying.

The unit's thesis meets a serious counter-argument. If no best practice can exist, it is odd that teleportation is the method most consumer experiences offer (Prithul, Adhanom and Folmer, 2021) and that redirected walking leads on spatial knowledge and preference in the head-to-head comparison (Langbehn, Lubos and Steinicke, 2018). On that reading a best practice does exist once the context is fixed, and the trade is a statement about the variety of contexts rather than the absence of guidance. The reply is that the context fix does the work, so the rule stays conditional on the task and the room.

The evidence behind each side is small: Usoh and colleagues (1999) replicated an earlier study, and Steinicke and colleagues (2010) measured thresholds with twelve participants.

Where this is heading

The direction is towards selection rules rather than a single recommendation. Dylong and Nickel (2025) bring mental workload into the comparison, turning a single verdict into a rule keyed to the task and the operator. Prithul, Adhanom and Folmer (2021) review the variants of teleportation. Boletsis (2017) supplies the shared typology the field lacked and names the gap: work concentrated on technology rather than experience. The need is not a new technique but a common outcome battery reporting presence, sickness, workload, spatial knowledge and physical space together. Grechkin and colleagues (2016) add one more: a threshold can shrink to a third when the estimation method changes, so any design rule should be quoted with the procedure that produced it.

Sources

• Usoh, M. et al. (1999). Walking > walking-in-place > flying, in virtual environments. SIGGRAPH '99. doi:10.1145/311535.311589 Replicated an earlier walking-in-place versus flying study and added real walking, which was significantly better on ease and higher on presence. Presence correlated most strongly with avatar association, so the result is entangled with body tracking.
• Whitton, M. et al. (2005). Comparing VE locomotion interfaces. IEEE Proceedings VR 2005. doi:10.1109/vr.2005.1492762 Had participants walk to wall targets under three locomotion and three visual conditions. Over 95 per cent of motion-path variance sat in three values, and a naturalness ordering appeared in the movement but not in the performance.
• Interrante, V., Ries, B. and Anderson, L. (2007). Seven League Boots: A New Metaphor for Augmented Locomotion through Moderately Large Scale Immersive Virtual Environments. IEEE Symposium on 3D User Interfaces. doi:10.1109/3dui.2007.340791 Proposed a method that amplifies only the motion component aligned with the intended direction. Participants overwhelmingly preferred it to flying, uniform-gain walking and periodic reorientation.
• Steinicke, F., Bruder, G., Jerald, J., Frenz, H. and Lappe, M. (2010). Estimation of Detection Thresholds for Redirected Walking Techniques. IEEE Transactions on Visualization and Computer Graphics. doi:10.1109/tvcg.2009.62 Measured redirected walking thresholds with twelve participants in three forced-choice experiments. Users tolerated 49 per cent more or 20 per cent less rotation, 14 per cent downscaling and 26 per cent upscaling of distance, and a 22 metre arc that felt straight.
• Suma, E. A., Lipps, Z., Finkelstein, S., Krum, D. M. and Bolas, M. (2012). Impossible Spaces: Maximizing Natural Walking in Virtual Environments with Self-Overlapping Architecture. IEEE Transactions on Visualization and Computer Graphics. doi:10.1109/tvcg.2012.47 Built self-overlapping architecture so a larger interior fits a smaller tracked area. Small rooms overlapped by up to 56 per cent before detection and rooms in a 9.14 metre square workspace by up to 31 per cent, and distances still seemed uncompressed.
• Hodgson, E. and Bachmann, E. (2013). Comparing Four Approaches to Generalized Redirected Walking: Simulation and Live User Data. IEEE Transactions on Visualization and Computer Graphics. doi:10.1109/tvcg.2013.28 Compared four redirected walking algorithms on simulated paths and live users, measuring distance from the tracking-area centre, wall contacts and redirection rate. Centre-steering out-performed the other three.
• Nabiyouni, M., Saktheeswaran, A., Bowman, D. A. and Karanth, A. (2015). Comparing the performance of natural, semi-natural, and non-natural locomotion techniques in virtual reality. 2015 IEEE Virtual Reality (VR). doi:10.1109/vr.2015.7223386 Compared real walking, a game controller and a semi-natural rotating-sphere device on task performance. The semi-natural technique was significantly slower and less accurate than both others.
• Bruder, G., Lubos, P. and Steinicke, F. (2015). Cognitive Resource Demands of Redirected Walking. IEEE Transactions on Visualization and Computer Graphics. doi:10.1109/tvcg.2015.2391864 Used a dual-task method, with verbal and spatial working memory tasks performed while walking under redirection. Redirection influenced both memory tasks, and the tasks influenced walking in return.
• Grechkin, T., Thomas, J., Azmandian, M., Bolas, M. and Suma, E. (2016). Revisiting detection thresholds for redirected walking: combining translation and curvature gains. ACM Symposium on Applied Perception. doi:10.1145/2931002.2931018 Re-measured the curvature threshold while translation gain was also applied. The threshold was unchanged, so the two can be combined, but the estimated detectable radius was 11.6 or 6.4 metres against the earlier 22 metres.
• Bozgeyikli, E., Raij, A., Katkoori, S. and Dubey, R. (2016). Point and teleport locomotion technique for virtual reality. Annual Symposium on Computer-Human Interaction in Play. doi:10.1145/2967934.2968105 Introduced point and teleport, in which the user aims at a destination and is moved there, compared with walking-in-place and a joystick with sixteen users. Users found it fun and user friendly, but a direction component degraded the experience.
• Boletsis, C. (2017). The New Era of Virtual Reality Locomotion: A Systematic Literature Review of Techniques and a Proposed Typology. Multimodal Technologies and Interaction. doi:10.3390/mti1040024 Reviewed empirical locomotion studies from 2014 to 2017, finding thirty-six articles describing seventy-three instances of eleven techniques, and proposed a four-type typology. It reports the research concentrated on technology rather than user experience.
• Langbehn, E., Lubos, P. and Steinicke, F. (2018). Evaluation of locomotion techniques for room-scale VR: joystick, teleportation, and redirected walking. Virtual Reality International Conference - Laval Virtual. doi:10.1145/3234253.3234291 Compared joystick, teleportation and redirected walking on spatial knowledge, effectiveness, sickness, presence and preference. Redirected walking gave the best spatial knowledge acquisition and, with teleportation, was preferred over the joystick, which caused more sickness.
• Prithul, A., Adhanom, I. B. and Folmer, E. (2021). Teleportation in Virtual Reality; A Mini-Review. Frontiers in Virtual Reality. doi:10.3389/frvir.2021.730792 Reviewed teleportation and its variants, reporting that it lets a user navigate beyond the tracking space with a low possibility of inducing sickness, and identifying the questions still open.
• Anderton, C., Creed, C., Sarcar, S. and Theil, A. (2024). From teleportation to climbing: a review of locomotion techniques in the most used commercial virtual reality applications. International Journal of Human-Computer Interaction. doi:10.1080/10447318.2024.2372151 Reviewed the locomotion options in 330 of the most used applications released between 2016 and 2023. Teleportation fell from 48 per cent of 2016 releases to 18 per cent in 2023, so the de facto standard claim no longer holds.
• Dylong, F. and Nickel, P. (2025). Virtual reality locomotion techniques and mental workload: an empirical investigation of point-teleport and directed-steering during virtual task performance. Virtual Reality. doi:10.1007/s10055-025-01290-2 Compared two variants each of point-teleport and directed-steering with twenty participants across three tasks. Point-teleport with a curved ray gave lower workload than the straight-ray variant, and both directed-steering variants impaired performance through visually induced sickness.