The NavBelt: A Computerized Travel Aid for the Blind

ABSTRACT

This project aims at the further development of a sophisticated computerized Electronic Travel Aid (ETA) for blind and visually impaired individuals. In its final prototype version, the device will consist of a belt with a small computer, ultrasonic and other sensors, and support electronics. Signals from these sensors will be processed by a unique algorithm and relayed to the user via headphones. This device, called NavBelt, will enable a blind user to walk safely and quickly through unknown, obstacle-cluttered environments.

In the photograph on the right, grad. student Shraga Shoval tries an early prototype of the NavBelt.


The NavBelt system provides acoustic feedback in two distinct modes of operation: the guidance mode and the image mode, explained below.

1. Guidance Mode
The Guidance Mode is a novel ETA concept for fast, safe travel. In this mode, either the direction of travel or the target location is known to the system. A single acoustic or tactile signal actively guides the user around obstacles in pursuit of the target direction. Preliminary experiments show that only a minimum of conscious effort is required by the user to follow this directional signal.

2. Image Mode
The Image Mode presents the user with an acoustic or tactile image (a 120o panorama) of the environment. Advanced statistical signal-processing algorithms compensate for sensor inaccuracies and provide the user with a more accurate and intuitive image than existing ETAs do.

The underlying technology of the NavBelt is based on our previously developed Obstacle Avoidance System (OAS) for mobile robots Ä a proven and mature technology considered by many experts to be one of the best mobile robot OAS in the nation.

The NavBelt ideas has been received with great enthusiasm by the Greater Detroit Society for the Blind and by numerous experts in the area of rehabilitation for the visually impaired. It should also be noted that in principle, the NavBelt could be interfaced with skin stimulators at a later time.


More details about the NavBelt Project can be found in papers 22, 38, 44, 52.

Project History

Johann Borenstein first conceived of the NavBelt system idea in 1989. He demonstrated the feasibility of the idea by riding blindfolded on a mobile robot, listening to acoustic directions generated by 24 ultrasonic sensors mounted on the robot. Using a joystick to direct the robot, travel speeds of 0.5 - 0.8 m/sec were demonstrated in obstacle cluttered environments. The NavBelt prototype was built in 1992 by grad student Shraga Shoval. A one-time $15K grant from the now dissolved Michigan Consortium for Enabling Technologies (MCET) paid for the prototype. Shraga Shoval was supported during his Ph.D. Thesis work by Johann Borenstein and Yoram Koren, who both co-chaired Shraga Shoval's Ph.D. committee. During that time Yoram Koren contributed many unique ideas aimed at improving the critical "man-machine interface." In December 1994 Johann Borenstein received a grant of $180K for three years from the Whitaker Foundation. This grant will be used to develop a small, lightweight portable Navbelt prototype suitable for testing with blind subjects.


This file last updated on 7/4/96 by Johann Borenstein

johannb@umich.edu