Artificial Neural Network-Based Shared Control for Smart Wheelchairs: A Fully-Manual Driving for the User

Abstract

Wheelchairs play an important role in regaining lost mobility and in the social reintegration of its users. However, users with quadriplegia yet have a few low cost solutions that meet their needs. The present work proposes a shared control strategy designed to operate together with discrete human-machine interfaces, where the user has few commands, and capable to integrate different types of sensors that can be attached to a commercial powered wheelchairs, without the necessity of localization. Simulations show that the proposed strategy provides a robust and safe navigation through daily environments and does not take away the navigation autonomy from the user, who performs a fully-manual driving.