The Car Attitude Tracking System Research

With the continuous improvement the quality of people's life and the development of social economy, the number of cars increase in a very high speed recently, traffic safety gradually become the focus of everyone's attention. Recent years more and more research focus on preventing dangerous driving behavior happing and reducing traffic accidents. For all the reasons mentioned above, there is an urgent need for a device capable of recording vehicle's trajectory and warning the driver when dangerous may happen, at the same time the records can be provided to the relevant departments for analysis. Existing cars with Attitude Tracking systems such as vehicle traveling data recorder is mainly used to record cars' speed or miles,however due to its complex structural and the high cost, it does not widely used in vehicles. While the Attitude Tracking System in this study mainly keep a watchful eye on the dangerous driving action and alter the driver in time, this system with the ability to study the dangerous driving behavior include abrupt braking, sharp turns and roll, and there is almost no time delay. The detail of each chapter as follows:Firstly, the structure of vehicle attitude tracking system is presented, and then an introduction about the data collection circuits is given. And select tri-axial accelerometer MMA8452 Q and angular velocity sensor ENC-03 MB as the detection sensor, both of them with the character of low cost and high precision.Then, on the base of referring relative papers and analyzing collected data, BP neural network was constructed, the numbers of layers of BP neural network, the transfer function and other related content was determined. After the process above,the data will be preprocessed, at the premise that ensuring the data accuracy,redundancy data was removed, then it is more easy to realize data compression, the final thing to do is extract the characteristics data, all of these process lay a solid foundation for the neural network recognition algorithm.Finally, preliminary gesture recognition was made according to the data, if the result shows it is belong to braking model or turn model, and then training the neural network to simulate the algorithm, realize identify vehicle's attitude. And the furtherverify the stability and accuracy of the system.