| Driving errors caused by excessive emotional fluctuations in the process,such as road rage,are one of the important factors leading to road traffic accidents,which may even cause serious casualties.In recent years,with the development of transportation and the improvement of science and technology,driving has become an inseparable part of people’s daily life.Therefore,the detection and early warning of driver emotion has become a hot spot in the research field.The existing detection work of emotion fluctuation is mainly based on visual and biological signal detection methods.However,the visual-based approach has problems of visual blocking or distortion,and the bio-signal-based approach has disadvantages such as intrusiveness,privacy invasion,and the use of equipment may also bring inconvenience or additional cost.On the contrary,wireless sensing technology based on WiFi signals has a series of advantages,such as non-line-of sight,non-intrusive,low cost,easy deployment,unrestricted by illumination conditions,strong expansibility,etc.,and is suitable for action recognition in vehicle space to detect drivers’ emotional fluctuations.From the perspective of the connection between WiFi signals and human movements,this thesis proposes a new WiFi signal-based driver emotion recognition framework,Widriver,to overcome the shortcomings of the existing methods.Wi Driver first designs the antenna position region through the Fresnel region to achieve the best signal acquisition effect.Secondly,by collecting Channel State Information(CSI)of drivers’ throttle and brake actions,emotion recognition coefficient is calculated.Based on recognition coefficient and Long short-term Memory network(LSTM)to identify emotion fluctuations.The experiment deployed Wi Driver in commercial WiFi infrastructure and evaluated its performance in real-world driving environments.The experimental results show that the average recognition rate of Wi Driver is 83.9% in the real scene. |
PDF Full Text Request