Improved classification and regression tree based omni-directional wheelchair control with eye movement
Junyou Yang, Guang Yang, Dianchun Bai, Zhicheng Liu, Long Zhou, Deguo Yang, Shuoyu Wang · 2016
Omni-directional wheelchairs are capable of providing great help for users with difficult walking as wish. In order to help the users with problems controlling wheelchairs with physical contact methods, this paper presents a set of algorithms including improved classification and regression tree (CART), cross validation separation(CVS) and relative angle model (RAM) to control a wheelchair with eye movements. The worldwide researches and some earlier work done about eye movement data processing were introduced which reveal the difficulty of applying eye movement data processing methods into motion control tasks. Afterwards, a kind of improved CART algorithm was proposed to efficiently decrease noisy data and extract useful information from raw data. And CVS along with RAM methods were presented to finish recognition tasks and improve the robustness of the system for nonstandard user generated scanpaths. In the last, an experiment was conducted to verify the effectiveness of the algorithms, and also the future work based on the results was proposed.