Users of virtual reality systems often need to navigate to distant parts of the virtual environment in order to perform their desired tasks. Unfortunately, physical space restrictions as well as tracker range limitations preclude the use of fully natural techniques for navigation through an infinite virtual environment. This necessitates the use of a locomotion interface, and the closer that interface matches the analogous real world actions, the easier it will be for the user. Unnatural techniques require cognitive effort on the part of the users. Many authors have attempted to address this problem by creating locomotion interfaces and techniques that more closely approximate real world counterparts to the extent possible. In addition to requiring these unnatural movements, current virtual reality systems are incapable of providing the high-fidelity sensory feedback used to guide real-world movements. This may cause users to resort to more cognitively demanding strategies.

There is a large body of research in the psychology domain regarding the structure of cognitive resources. In particular, Baddeley's multi-component model of working memory describes a separation between the resources used for verbal and non-verbal storage and processing. It is likely that semi-natural locomotion techniques require some of these resources, which will then be unavailable for concurrent tasks. A pair of studies was conducted, investigating the cognitive resource requirements of several atomic locomotion movements by manipulating the user interface and field of view. The results indicate that semi-natural locomotion interfaces generally require a user's spatial cognitive resources. Based on the conclusions from the working memory studies, an adaptive system was designed that can learn how to adjust parameters of the locomotion technique according to a user's present cognitive task load.