Study on the Technique of Information Fusionapplied to Enbedded Driver Fatigue Detection
|School||Zhejiang University of Technology|
|Course||Detection Technology and Automation|
|Keywords||Information fusion Driver fatigue recognition GPS Driving time Driving location TS fuzzy neural network|
Road traffic accidents occur frequently, which caused serious personal injury and economic losses. According to statistics, the driver fatigue is an important cause of road traffic accidents, which has attracted worldwide attention. Thus, the real-time detection of the driver fatigue has the vital practical significance in reducing the accidents caused by driver fatigue. Since the embedded system has features of easy-deployed, high reliability and low-consuming, the embedded drive fatigue detection system is becoming the main trend of future drive fatigue detection system. In this paper, it is made that a conclusion and research on the drive fatigue detection technology in domestic and abroad, and present extracting many effective driver fatigue characters from satellite positioning information and non-image-based, finally using the fuzzy logic and the artificial neural networks technology to get fusion detection and provide the early warning about the state of fatigue, in view of the limitation of detecting by single feature with image processing technique. The major work and creative results are presented as follows:Firstly, the basic principles of information fusion and its functional structure model are studied and summarized. The features and application of common information fusion algorithms are analyzed and compared, meanwhile the issues of how to select fusion algorithm in the practical application are to be considered.Secondly, the evaluation method of fatigue and driver fatigue are studied, and focusing on the comparison of driving fatigue evaluation methods based on visual and based on behavioral characteristics of the vehicle. The impact of driver fatigue by the driving time and driving location has been studied in the paper and it is proposed that a driving fatigue comprehensive evaluation methods based on driving time, driving location and behavioral characteristics of the vehicle.Thirdly, study on the extraction of driving time and driving location. It is proposed a new method of extracting driving time and driving location from satellite positioning and digital map information, then it is described that the implementation process to extract the real time and location information from GPS(Global Positioning System) information in detail.Fourthly, study on the corresponding connection of inconsistent situation of driving direction and driver reaction, the steering wheel action status and continuous driving time with the driving fatigue, and the new feature parameters such as inconsistent percentage of direction changing and driver reaction(DC&DR), continuous fixed time percentage of steering wheel(CPSW), percentage of continuous driving time(PCDT) are extracted from behavioral characteristics of the vehicle to get comprehensive integration parameters judgments.Finally, TS(Takagi-Sugeno) fuzzy neural network is introduced to detect the driver fatigue. The five fatigue characteristic parameters named driving time, driving location, DC&DR, CPSW and PCDT, which are extracted from GPS data and driving behavior of vehicle are fused by the TS fuzzy neural network and the output is just corresponded to the classification of the fatigue, compared these value with the PVT(Psychomotor Vigilance Task) value, the degree of the driver fatigue is decided and then the corresponding warning degree is confirmed.The results of the experiment have indicated that the proposed method is more efficient in real-time fatigue detection.