Terrain Characterization Using Modified RANSAC Analysis of Human Gait Data
One of the many lasting side effects of a stroke can be foot drop, or an inability to dorsiflex the foot. In order to remedy this, many people wear an ankle-foot orthotic (AFO) post-stroke. One of the many troubles these individuals face is in dealing with obstacles such as stairs and ramps, because the AFO limits the plantarflexion that is natural in navigating these obstacles [1,2]. The end goal of this research is to create an active AFO that adapts to changing ground terrain, providing a more natural gait pattern. This paper presents the first part of this work: a means for identifying terrain in order to control an AFO. This has been accomplished using an infrared (IR) range sensor attached to the lower leg, used to measure the surface profile of the ground just ahead of a test subject. Using a modified RANSAC technique to fit experimental gait data, standardized gait profiles for different terrain have been quantified and shown to be reproducible, indicating the utility of the technique for terrain identification and AFO control.