Research On Motor Imagery Brain Computer Interface For Intelligent Assisted Driving

At present,the decisions made by smart vehicles including unmanned vehicles often make drivers feel awkward and aggravate their nervousness.Moreover,existing smart vehicles reduce driver's driving participation,making them vulnerable to safety hazards such as drowsiness or distraction.At the same time,such vehicles lack effective interaction paths with disabled drivers,so they cannot obtain the driving intentions of disabled drivers and cannot deliver them to their destinations.If the brain computer interface(BCI)technology can be used to translate the electroencephalogram(EEG)signals to obtain the driving intentions of the limb-disordered drivers and control the vehicles,the range of movement of the disabled people will be greatly expanded,and the driving comfort will also be improved.The BCI system can establish a direct path between the human brain and a computer or other external devices.With the help of the BCI,people can interact with the outside world without physical action.As an important branch of the system,the motor imagery(MI)BCI system takes the spontaneous EEG signal as the signal source and can be obtained without any external stimulation and intrusion,so it has received extensive attention.One of the key aspects of the MI BCI system is the feature extraction and accurate classification of EEG.This paper focuses on the research of MI BCI in the field of intelligent assisted driving.In order to improve the recognition rate MI BCI and make MI BCI applied in intelligent assisted driving,the following work has been carried out:(1)The characteristics of MI EEG signals is studied based on the mechanism of the produce of MI EEG signals.Then the paper uses variational mode decomposition(VMD)and Hilbert transform to construct a single-channel framework of MI EEG signals.The framework uses VMD to extract the best high-dimensional time-frequency features to classify,which avoids the omission of information caused by choosing the optimal period and frequency band manually.(2)The paper uses multivariate empirical mode decomposition(MEMD)to construct a multi-channel framework of MI EEG signals.While extracting time-frequency characteristics after processing EEG signals with MEMD and Hilbert transform,the framework extracts nonlinear dynamic coupling feature,phase coupling feature and frequency coupling feature to improve the recognition rate of MI.(3)According to the proposed method,the proposed intelligent assisted driving platform is built.The correctness of the proposed method and the feasibility of the MI BCI in intelligent assisted driving are verified by experimental data and offline competition data.