Research Of The Method For Driver Identification Based On Smartphones

With the rapid development and growing popularization of hardware and software.Smartphones play an increasingly important role in people’s lives.In addition to the basic communication function,smartphones offer many convenient services for entertainment,office,society and finance.However,smartphones are also double-edged sword,because smartphones also bring some problems when they are convenient for people’s lives.The traffic safety problems caused by smartphones are particularly prominent.According to the relevant statistics,the proportion of traffic accidents caused by smartphones is increasing year by year.Most of these accidents are due to the driver’s attention being disturbed by smartphones.The problem that how to reduce the interference of smartphones to drivers is an urgent one.With the continuous development of smartphone hardware,the sensors make smartphones have a stronger sense of the outside world.Smartphones will have the ability of context awareness through the collaborative work of modules such as sensors,CPU and communication component.The ability to perceive the context of the smartphone lays foundations for solving the problem of reducing the interference of the smartphone to the driver.When a smartphone knows that a mobile user is a driver during driving,the smartphone will develop strategies to reduce the interference of the phone to the driver while does not has effect on other passengers.In this thesis,we propose a scheme about driver recognition based on smartphones.In our scheme,we will only use one smartphone and the sensors of the smartphone to identify whether the smartphone user is a driver during driving.We will not need any external sensors,networks or computing equipment.The basic idea of our scheme is as follows: We identify drivers and passengers by looking for specific actions that the driver has shown during driving.We will find specific actions with the smartphone’s awareness ability.When we have known those specific actions,we can infer whether the smartphone user is driver.The focus of this paper is as follows:(1)We look for some reasonable and reliable specific actions during driving to distinguish between the driver and the passengers.Both the driver and the passengers show meaningful and fixed behaviors during driving.For example,the driver will turn the steering wheel but the passenger will not.Extensive experiments demonstrate that we can distinguish them using five types of the actions: the action of moving leg when you get on the car,the action of fastening seat belt,the action of hitting the brake or accelerator,the action of steering wheel and the car runs over the uneven roads.(2)We focus on designing effective driver identification solution.How to effectively capture the specific actions to distinguish between the driver and the passengers? How to infer whether the smartphone user is a driver when we analyze the specific actions.In this thesis,support vector machine(SVM)technology will be used to identify the specific action,and we design a solution to infer whether the smartphone user is a driver or not.(3)We implement and test the driver recognition solution based on Android operating system.We test the proposed solution in the real driving environment.The experimental results demonstrate that the accuracy of each specific action can reach at least 85%,and the correct rate of driver identification is at least 90% during driving.They show that the driver recognition scheme proposed in this paper is feasible and effective.