Using Inertial Data for Locomotion in Virtual Environments

This technical report describes a simple method for using inertial data to facilitate locomotion in virtual environments. The means of locomotion for virtual environments contributes greatly to the users’ sense of presence in the environment. Although walking-in-place is not so realistic as real walking, it is a very cost-effective alternative. Using a simple version of this algorithm both Type I and Type II errors were decreased compared to a previous walking-in-place algorithm. Type I errors occur when the algorithm detects a step when none occurred. Type II errors occur when the algorithm does not detect a step when one did occur. Type I errors decreased from 3% to 1% and Type II errors decreased from 32% to 11%. The more complex version of this algorithm promises even better results. Introduction The means of locomotion in virtual environments plays an important role in the user’s sense of presence [Slater 1994]. Additionally, walking-in-place (virtual walking) as a means of locomotion in virtual environments is a cost-effective alternative to other forms of locomotion. The cost for equipment and lab space is much less than other techniques such as wide-area tracking. Unfortunately, the innaccuracy of many virtual walking methods is problematic. Previous algorithms based on a neural network had Type I and Type II errors as 3% and 32% respectively [Usoh, 1999]. However, the use of inertial information to detect footsteps presents a potentially cheap, accurate alternative to current means of locomotion. The technique described in this report can provide good results at a fraction of the cost of other techniques. In this case inertial data is gathered by a Crossbow Accelerometer and a National Data Acquisition board. The accelerometer is fastened to the head mounted display (HMD). This location was chosen because of the existing attachment points and cables that are used for the HMD. Other locations such as the foot or back could also be used, but these possibilities have not been studied.