Online Learning Shared Control System of Electric Wheelchair for Persons with Severe Disabilities

Abstract

Persons with severe disabilities experience difficulties in operating an Electric Wheelchair (EW) due to their lack of muscle power and the deformation of their hands. Even special input devices are developed to meet their physical limits, there is no guarantee that they can operate the EW correctly as they intend. To deal with this, EW needs to add more or less intelligence. Unlike autonomous driving, shared control can not only improve the operation but also keep users’ residual physical function and improve their enthusiasm. However, it is always a difficult problem to decide the control weights between user and machine to meet the requirements that making full use of users’ operating abilities and giving assistance when necessary. In this paper, we propose a novel online learning shared central system that can adjust the central weights considering users' motivation, safety and social adaptabilities by online evaluating users’ operating ability. Simulation results show the validity of the proposed method.