The Research And Implementation Of The Perception Algorithm Based On Contextual Perception

As the comfort requirements of passengers when they are taking vehicle ride gradually increase, it has been determined as an essential issue that Intelligent Transport Systems(ITS) confronts and affects passengers’ satisfaction. During steering, the driving behavior and style as well as the road condition is the main factor affecting the riding comfort. Practice has proved that the research on riding perception plays a positive role in enhancing the riding comfort. To detect the riding perception on every street effectively and choose a routine with generally greater driving comforts, however, is an effective approach to improve the riding perception. Currently, the hardware of the smart phone develops ceaselessly, more and more sensors such as the acceleration transducer, gyroscope and gravity sensor are on mobile phones. What a pity that they have not been best used of.This thesis is based on the smart phone apparatus and considers participatory perception as the kernel of the conception. By using of adopt cross-sourcing modes and the introduction of riding comfort, we proposes a context-aware sensing identification system for detecting and assessing the ride perception. Based on the accelerometer and magnetometer, the system firstly presents a coordinate transformation system. Then It converts triaxial acceleration sensor data on the phone to the acceleration data of the vehicle, thus it eliminates the impact on acceleration sensor signal introduced by the arbitrariness of mobile phone being kept at random position and direction on the car. This system will be further divided into a horizontal perception level perception and vertical coordinate system which were checked and evaluated separately. Ride perception in the vertical direction may be affected by road conditions, such as potholes, steep slopes. On the other side, ride perception in the horizontal direction may be affected by a number of contextual information, such as the driver’s style of driving behavior(speeding forward, braking, cornering, etc.).According to the evaluation of human vibration in ISO2631 ride perception, we propose and design an evaluation mechanism to evaluate the riding perception during driving period, which is closer to the genuine body feeling of riding.Finally, the system cannot only decide the driving styles of the drivers according to their driving events, but also give alarms when dangerous driving behavior appears. The whole process is done on the intelligent terminals, which is low cost, strong real-time performance, high accuracy and easy to use.