The Navchair: an Assistive Navigation System for Wheelchairs Based upon Mobile Robot Obstacle Avoidance


(Note: this Abstract and Introduction adapted from Paper #50)

The NavChair assistive navigation system is being developed to meet the needs of multiply handicapped people who are unable to operate available wheelchair systems. The NavChair Project was conceived as an application of mobile robot obstacle avoidance to a power wheelchair. During the course of this project, the Vector Field Histogram (VFH) (see Paper16,17 ) method has been adapted for use in human-machine systems and the shortcomings of the wheelchair platform have been overcome.

The NavChair, fully equipped with 12 ultrasonic sensors and an onboard computer.


The NavChair assistive navigation system is being designed to improve the mobility and safety of people who have impairments that limit their ability to operate a power wheelchair. The NavChair control system is being built to avoid obstacles, follow walls, and travel safely in cluttered environments. This paper outlines research into the application of robotic obstacle avoidance to this power wheelchair assistive navigation system.

The NavChair is a human-machine system in which the machine must share control with the user. VFH obstacle avoidance modifys the user's input command to achieve safe travel. This approach allows the user effective control while overriding unsafe maneuvers. Two types of problems were encountered in the application of the VFH method to a power wheelchair system. First, the power base is significantly different than typical mobile robots. For example, the pneumatic tires, wheel slippage, and loose drive train make dead-reckoning accuracy an order of magnitude worse in power wheelchairs than in most mobile robots. In addition, the geometry and kinematics of the wheelchair are significantly more complicated than those of most mobile robots. The second type of difficulty is related to the application of obstacle avoidance to a human-machine system. Safe travel is only one of several requirement for the NavChair controller. In addition, the user must feel safe and in control of the wheelchair; the system's reaction to input must be intuitive enough to inspire confidence and allow understanding; and the wheelchair's motion must be smooth and comfortable. Existing obstacle avoidance methods were generally developed for autonomous mobile robots under a significantly different set of erformance goals.

The NavChair project has been largely successful in overcoming these difficulties. At present, blindfolded subjects can travel in unstructured indoor environments at high average speeds (up to 0.8 m/sec) without any collisions. These results are demonstrated in representative indoor environments that include obstacles such as smooth walls, 8 mm (1/4 inch) poles and office furniture. In addition, preliminary results suggest that users perceive the system as safe, comfortable, and intuitive. However, the NavChair achieves safe travel by being conservative, which means that it cannot achieve all possible 'safe' behavior{s}. For example, VFH obstacle avoidance does not allow the NavChair to {pass through some doors, even though they may be wide enough for passage}.

More details about the NavChair project can be found in papers 23, 37, 41, 42, 50

Project History

The idea of the NavChair was first conceived by Simon Levine (Director of Physical Rehabilitation at the University of Michigan Hospital) after being presented with a demo of the VFH Obstacle Avoidance System for Mobile Robots, developed by Johann Borenstein and Yoram Koren. The project got started in 1991 under a 3-year $330K grant from the Veteran's administration, with Simon Levine as the Principal Invetstigator. Grad. student David Bell and Senior Computer Engineer Lincoln Jaros first implemented the VFH obstacle avoidance method and showed successful navigation. In subsequent work David Bell made substantial improvements to the system, resulting robust handling of difficulties, such as narrow doorway passage and specular environments.
This file last updated on 7/4/96 by Johann Borenstein