The Research Of Electric Wheelchair Based On Consciousness Eyes Movement

Human is connected with computer or other electronic equipment byHuman-computer interaction technology which is called HCI for short and it canreflect human’s will. With the progress of science and technology, the means of HCIhave become more diversified, and HCI has brought more convenience to people’slife. However, the primary means of HCI must be operated by the user’s body, and itis difficult for the people with movement disorders or the aged to control, sodeveloping a new interactive tool is very necessary. Currently, the applicationresearches on bio-electricity signal become more widely, and bio-electricity signalapplied to the field of HCI has been a hot spot of researches.Electro-Oculogram(EOG), is a kind of bio-electricity signals because of thepotential difference between the cornea and retina of the eyes. With the rotation of theeyes, the signal will be changed, moreover the signal of various modes of eyemovement has distinguishable feature. So it is a feasible method theoretically that themode of eye movement can be identified through the EOG signal, and then convertedto the commands to control external devices.Supported by the national natural science fund project, and according toproblems of the processing technology of EOG signals and combined the task request,this paper designs a system to control the electrical wheelchair based on the consciouseyed movement. The system concludes the collection of nine kinds of EOG signals ofeyes movement and de-nosing pre-processing, feature extraction and detection, finallythe effectiveness and real-time performance of the system is verified online. Thespecific work during the research and main innovations during this study is listed asfollows:(1)The acquisition scheme is designed according to the characteristics of theEOG signals: choosing a bipolar lead mode to collect the signals of the horizontal andvertical channels by the equipment, to acquire EOG signals of nine kinds of eyesmovements, respectively are blink twice, glance to the left, right, up, down, left-up,left-down, right-up and right-down direction. In the per-processing stage, the signalsare de-noised by Butterworth low-pass filter and then adopting the short-term energyendpoint detection scheme to acquire valid period of EOG signals. And the signals are normalized to eliminate differences among different subject. Lastly the signals areremained at200sampling points by the one-dimensional interpolation.(2)In the feature extraction stage, proposing the feature of EOG signals based onthe waveform, then using differential linear prediction coefficient and wavelettransform to extract the features of EOG signals. The characteristics of two channelsare combined as the feature vectors.(3)In the pattern classification stage, three classifiers are designed, respectivelythe S_Kohonen network, support vector machine and fuzzy support vector machine.Firstly, the EOG signals are directly classified by the EOG signals, taking the lowfrequency coefficients of wavelet decomposition as feature, the recognition rate canreach more than94%. Using support vector machine and fuzzy support vectormachine and the low frequency coefficients of wavelet decomposition, the accuracyrate reaches more than96%and97%respectively. By comparing the experimentresults, selects the low frequency coefficients of wavelet decomposition as feature andfuzzy support vector machine as classifier which provide theoretical and experimentalfoundation for the wheelchair control system.(4)An online electric wheelchair controlled by consciousness eyes movement isdesigned, it can be distinguished that following EOG signals are conscious orunconscious by blink twice. In this control system, only the EOG signals producedafter blink twice in1s can be converted to effective control commands, otherwise it isregarded as invalid instruct and wheelchair will not do any action, so this scheme caneffectively avoid the interference of awareness eye movement. Because of variouspatterns of eyes movement, three kinds of speed are added to the advance directive.The practical experiment shows that the scheme is feasible.