OBSTACLE SEGMENTATION AND WARNING SYSTEM FOR VISUALLY IMPAIRED PEOPLE

Computer Engineering

This work describes a wearable obstacle object segmentation and warning system that can improve the dynamism and safety of blind and visually impaired people, particularly in unknown environments. The system uses Simultaneous Localization and Mapping and semantic path planning to complete localization and navigation tasks while collaborating with the visually impaired user. The RGB-D sensor camera detects and identify the location obstacles, and then sends information about the obstacles to visually impaired people via various modalities such as voice, tactile, and vibration. The path and motion guidance are generated to direct the user to their desired destination. The hardware of the system is consists of RGB-D camera, a Raspberry Pi module, and an eyeglasses. The software modules are implemented in MATLAB and OpenCV, and the navigation system is tested with blindfolded observers. The system employs computer image processing and motion planning strategies to (1) determine the accessible space, (2) construct an easy and safe movement path in space, and (3) identify and recognized particular kinds of objects, such as an empty chair. All of this information is transmitted to those who are wearing the device via vibration. We present findings from user studies on lower and high-level activities such as avoiding collisions in a maze, locating a chair, and navigating an overcrowded space while avoiding people.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.