A wearable navigation system for augmenting guide dog object detection

Abstract

In this thesis, I discuss the design, construction and evaluation of a wearable object detection system for blind users to avoid collisions. Blind people rely upon tools such as guide dogs to navigate the world around them. However, even guide dogs are prone to errors that leave their owners prone to unexpected collisions with objects, such as low-hanging branches and street signs, that may cause bodily harm. This work introduces a wearable device that provides object detection capabilities to blind people with guide dogs. The device contains a camera, an embedded computer and a vibration motor to provide feedback when an obstacle is detected. The system uses computer vision techniques to determine whether there are objects in the path of the user, as well as the distance of said objects. A series of experiments was conducted with guide dog users traversing a path of obstacles while using the device to determine the effectiveness of the device at allowing users to avoid collisions.