Development of a Method to Identify Walking Pattern and Performance in a Virtual Reality Obstacle Course

The purpose of this project was to develop a method to assess walking performance of older adults on a virtual reality (VR) obstacle course. Thirty older adults participated in a 6 week treadmill training program augmented by VR aimed to decrease fall risk. The VR system consisted of an augmented reality simulation and a camera based motion capture (Kinect) which monitored the subject’s feet movements walking on the treadmill while negotiating different types of virtual obstacles. We developed unique algorithms based on visual analysis, Fourier methods, and statistical tests to help characterize the pattern of obstacle negotiation and determine the effects of training on performance. The method decreased the signal to noise ratio of the obtained data and allowed determining each walking step with high sensitivity and accuracy. Common spatio-temporal measures of gait such as step length were extracted from 3 phases of the walk: (1) preparation of obstacle negotiation, (2) stepping over the obstacle, and (3) gait recovery after the obstacle. Our findings indicate improvement in step length throughout the training. Effects of training on obstacle negotiation are in progress.