Predicting Wheelchair Stability While Crossing a Curb Using RGB-Depth Vision

Abstract: Handicapped individuals often rely heavily on various assistive technologies including wheelchairs and the purpose of these technologies is to enable greater levels of independence for the user. In the development of autonomous wheelchairs, it is imperative that the wheelchair maintains appropriate stability for the user in an outdoor urban environment. This paper proposes an RGB-Depth based perception algorithm for 3D mapping of the environment in addition to dynamic modeling of the wheelchair for stability analysis and prediction. We utilize RTAB Mapping in combination with Poisson Reconstruction that produced triangular mesh from which an accurate prediction of the stability of the wheelchair can be made based on the normals and the critical angle calculated from the dynamic model of the wheelchair.