Depth map based reactive planning to aid in navigation for visually challenged

Abstract

Visual information is essential for navigation oriented tasks, but visually challenged individuals face difficulties in performing these tasks since they lack sufficient information regarding their surrounding environment. In the literature, many vision and non-vision based aids have been proposed to help the blind for navigation and perception in general. The proposed system aims to a provide reactive planning method using a depth map acquired from a vision system, to assist visually challenged individuals in achieving collision-free navigation in an indoor environment. Understanding the environment in front involves two phases. Firstly, classifying the obstacles as static or dynamic will help us identify the obstacles that require a quick response. Secondly, the velocity of the obstacles is estimated from the depth map which gives information whether they are approaching or moving away. Once understanding the environment in front is carried out regarding its geometry, navigation planning is done to guide the user who holds the differential drive robot attached to the walking stick.